v2.17.1

Compare Source

Patch Changes

• #​17578 7cc0b00 Thanks @​renovate! - Fixes a case where rich content diagnostics and snippet edits would not work

v19.3.0

Compare Source

New React Features

• <ViewTransition />: Adds <ViewTransition /> and addTransitionType APIs to power View Transition animations in React (@​sebmarkbage, @​jackpope, @​gaearon: #​31975, #​31987, #​31996, #​31999, #​32001, #​32002, #​32028, #​32029, #​32031, #​32034, #​32038, #​32041, #​32050, #​32090, #​32105, #​32254, #​32379, #​32422, #​32462, #​32540, #​32545, #​32585, #​32599, #​32611, #​32612, #​32617, #​32651, #​32653, #​32656, #​32664, #​32699, #​32723, #​32734, #​32751, #​32752, #​32760, #​32761, #​32764, #​32772, #​32790, #​32819, #​32820, #​32822, #​32833, #​32849, #​33094, #​33191, #​33200, #​33206, #​33293, #​33330, #​33331, #​33332, #​33357, #​33362, #​33433, #​33576, #​34374, #​34450, #​34481, #​34500, #​34502, #​34510, #​34511, #​34539, #​35567, #​35564, #​35485, #​35380, #​35063, #​35060, #​34676, #​36917, #​35337, #​35520)
• Fragment Refs: Add Refs to <Fragment /> to support composable platform behavior (@​jackpope, @​sebmarkbage, @​eps1lon, @​Dhakshin2007, @​chirokas, @​teamleaderleo, @​fallintoplace: #​32465, #​32613, #​32619, #​32654, #​32660, #​32682, #​32722, #​32813, #​32814, #​33056, #​33058, #​33093, #​34069, #​34103, #​34544, #​34545, #​37062, #​37061, #​37060, #​36047, #​36010, #​35642, #​35641, #​35637, #​35630, #​34935, #​37457, #​37408, #​37326, #​37251, #​37171, #​37169, #​37168, #​37167, #​37166, #​37165, #​37164, #​37163, #​37162, #​37161, #​37160, #​37125, #​37063)

New React DOM Features

• browser(): a new react-dom API that returns a usable which errors during server rendering and resolves in the browser. use(browser()) inside a <Suspense> boundary marks a subtree as browser-only without reporting a recoverable error (@​gnoff: #​37143, #​37241)
  • Added an onBrowserBailout option to the react-dom/server APIs to observe when a subtree defers to the browser (@​gnoff #​37193)

Notable changes

• Enable Trusted Types API integration (@​rickhanlonii #​35816)
• Transitions now render independently instead of being entangled into a single render, so a slow transition no longer holds up unrelated ones (@​acdlite #​37290)
• Added a DEV-only warning when a component appears to have been unblocked by calling use() conditionally (@​hoxyq, @​eps1lon: #​37104, #​37203, #​37491)

All Changes

React

• Fast Refresh Fixes
  • Fix Fast Refresh to find and remount edits to components wrapped behind lazy() (@​sophiebits #​36965)
  • Fix Fast Refresh so edits to a memo() comparison function take effect (@​sophiebits #​36964)
  • Fix Fast Refresh crash when an edit changes the kind of a component's type (@​sophiebits #​36963)
  • Unify hot reload type resolution for Fast Refresh (@​sophiebits #​36962)
  • Fix Fast Refresh to remount correctly when an edit changes the component kind (@​sophiebits #​36950)
  • Double invoke effects in StrictMode after Fast Refresh (@​eps1lon #​35962)
• Performance Track Fixes
  • Prevent crash when accessing $$typeof in Performance Tracks (@​eps1lon #​35679)
  • Handle non-string function names in Performance Tracks (@​eps1lon #​35659)
  • Use minus (-) instead of en dash for removed props in Performance Tracks (@​eps1lon #​35649)
  • Handle arrays with bigints in deep objects in Performance Tracks (@​eps1lon #​35648)
  • Don't enumerate typed array props in Performance Tracks in DEV (@​UditDewan #​36913)
  • Bail out of diffing wide objects and arrays in Performance Tracks (@​eps1lon #​34742)
  • Clear potentially large performance measures in DEV (@​hoxyq #​34803)
  • Fix missing else branch for renders with no props change in Performance Tracks (@​hoxyq #​34837)
• Activity Fixes
  • Fix useSyncExternalStore missing store mutations that happened while an Activity tree was hidden (@​sophiebits #​36947)
  • Hide portal contents when an Activity is hidden (@​acdlite #​35091)
  • Prevent metadata hoisting in hidden <Activity> trees (@​ronnakamoto #​34983)
  • Prevent errors thrown inside a hidden Activity from escaping to the visible UI (@​acdlite #​35074)
  • Don't unhide a node if a direct parent Offscreen is still hidden (@​sebmarkbage #​34821)
  • Don't show internal <Offscreen> component in error messages (@​rickhanlonii #​35763)
• Warn in DEV when a component appears to have been unblocked by a conditional use() (@​hoxyq, @​eps1lon: #​37104, #​37203, #​37491)
• Render transitions independently instead of entangling them into a single render (@​acdlite #​37290)
• Fix hang when updating a dehydrated boundary inside a hidden tree (@​gaearon #​37135)
• Don't reacquire Host Singletons during dev effect validation (@​gnoff #​37113)
• Only remove properties from Host Singletons on release (@​gnoff #​37112)
• Don't invoke effects on moved children in StrictMode (@​sophiebits #​36948)
• Throw a clearer error when a rejected Promise is incorrectly instrumented (@​eps1lon #​36328)
• Fix context propagation into Suspense fallbacks (@​acdlite #​36160)
• Fix context propagation through suspended Suspense boundaries (@​acdlite #​35839)
• Fix useDeferredValue getting stuck (@​gaearon #​36134)
• Update useActionState error message to say "action state" instead of "form state" (@​rickhanlonii #​35790)
• Don't warn when rendering data block scripts (@​eps1lon #​35953)
• Treat fibers with alternates as mounted in getNearestMountedFiber (@​sebmarkbage #​35578)
• Always instrument the lazy initializer thenable (@​gnoff #​35521)
• Annotate cascading updates for DevTools performance tracks in DEV (@​huntie #​34961)
• Resolve the .default export of a React.lazy as the canonical value (@​sebmarkbage #​34906)
• Limit owner stacks to 10 frames (@​eps1lon #​34864)
• Ensure useEffectEvent reads latest values in forwardRef and memo() components (@​eps1lon #​34831)

React DOM

• Copy source onto the synthetic toggle event (@​lazerg #​37389)
• Update HTML parser rules for the new <select> parser (@​josepharhar #​34804)
• Update defaultValue for type="number" inputs to match other input types (@​sophiebits #​36980)
• Avoid setting innerHTML when it hasn't changed (@​sophiebits #​36949)
• Forward the nonce option in ReactDOM.preloadModule() (@​UditDewan #​36851)
• Support fetchPriority for module resources (@​TheAlexLichter #​36835)
• Track stylesheet preloads when explicitly preloaded to avoid unnecessary suspension (@​gnoff #​36386)
• Suggest correct casing for the misspelled credentialless iframe attribute (@​vmx906 #​36322)
• Add credentialless as a recognized boolean attribute for iframes (@​vmx906 #​36148)
• Prevent reentrant finishedTask from calling completeAll multiple times (@​gnoff #​36287)
• Update the CSS shorthand property list (@​kidonng #​35636)
• Support the maskType SVG prop (@​sleitor #​35921)
• Add support for onFullscreenChange and onFullscreenError events (@​chirokas #​34621)
• Include submitter in submit events (@​eps1lon #​35590)
• Fire onReset when automatically resetting forms (@​eps1lon #​35176)
• Use the FormData submitter parameter (@​jenseng #​29028)
• Fix form status reset when component state is updated (@​eps1lon #​34075)
• Batch updates from resize events until the next frame (@​eps1lon #​35117)

React Server

• Add nonce to rendered import maps (@​mattcosta7 #​37339)
• Add an onBrowserBailout option to observe subtrees that defer to the browser (@​gnoff #​37193)
• Stop the caller's signal from retaining a finished render (@​unstubbable #​37315)
• Construct the render lifetime controller only when it is needed (@​unstubbable #​37357)
• Fix react-dom/server hanging on Deno (@​fraidev #​35235)
• Extend stack overflow recovery to retries in SSR (@​jackpope #​36977)
• Fix false-positive hydration mismatch on nonce attributes (@​MaxwellCohen #​37030)
• Guard shell-error callbacks instead of nulling them out on the request (@​eps1lon #​36916)
• Stop firing onAllReady after the shell errored (@​eps1lon #​36903)
• Treat an incomplete tree as an error during error recovery (@​acdlite #​36911)
• Fix non-script modulepreload tracking (@​fallintoplace #​36564)
• Finalize postponed nextSegmentId after the prelude flush (@​gaojude #​36779)
• Preserve referrerPolicy in image preload headers (@​ulises-jeremias #​36803)
• Allow pending work to specialize abort reasons (@​gnoff #​36586)
• Abort tasks that suspend after aborting during render (@​gnoff #​36585)
• Finish abort in a scheduled task (@​gnoff #​36580)
• Fix aborts during resumed rendering (@​gnoff #​36584)
• Track abort state on the request (@​gnoff #​36583)
• Model external work scheduling explicitly in the fb bundle (@​gnoff #​36576)
• Continue reporting aborted task errors after a fatal abort (@​gnoff #​36575)
• Disallow abort reentrancy (@​gnoff #​36574)
• Stop transpiling computed property names (@​eps1lon #​36542)
• Fix crash when capturing the callsite of a stalled use() of a rejected Flight chunk (@​lubieowoce #​36544)
• Double invoke effects in Strict Mode during hydration (@​eps1lon #​35961)
• Add additional task reentrancy protections in Fizz (@​gnoff #​36291)
• Don't outline Suspense boundaries with suspensey CSS during shell flush (@​gnoff #​35824)
• Push a stalled use() to the owner stack and debug task (@​eps1lon #​35226)
• Fix useId tracking on replay (@​gnoff #​35518)
• Correctly handle replaying when hydrating (@​gnoff #​35494)
• Simplify the createSuspenseBoundary path (@​sebmarkbage #​35087)
• Unify preamble-only fields to save a field on Suspense boundaries (@​sebmarkbage #​35068)
• Unify prerender-only fields to save a field on Suspense boundaries (@​sebmarkbage #​35067)
• Push halted awaits to the owner stack for late-arriving I/O info (@​unstubbable #​35019)
• Include all Node.js APIs in the Bun entrypoint for react-dom/server (@​RiskyMH #​34193)
• Detach a Suspense boundary after flushing its segment (@​gnoff #​34694)

React Server Components

• Keep an element pending while a referenced row is resolving (@​unstubbable #​37542)
• Fix RangeError from exponential debug info growth (@​unstubbable #​37481)
• Abort the cache signal when debug objects are retained (@​unstubbable #​37342)
• Transfer key validation of lazy nodes when they are unwrapped (@​unstubbable #​37258)
• Add pending_weak to the Flight thenable protocol (@​acdlite #​37154)
• Outline and dedupe repeated strings (@​gaearon #​37147)
• Port Reply Server traversal guards to the Flight Client (@​eps1lon #​37144)
• Define Flight chunk .then with Object.defineProperty (@​eps1lon #​37109)
• Support Turbopack's experimental array format for chunks (@​sampoder #​37095)
• Limit fake JSX call site stacks to 10 frames (@​timneutkens #​37086)
• Improve performance when decoding Flight replies (@​gnoff #​37090)
• Recognize Node 22+ V8 frames for Promise statics in debug info (@​gaearon #​36969)
• Tailor the warning for binary data serialized through toJSON (Node Buffer) (@​unstubbable #​36873)
• Resolve models before JSON.stringify for faster serialization (@​mhart #​36795)
• Prune debug info when chunks error (@​gnoff #​36782)
• Avoid main-thread stalls from large debug strings (@​unstubbable #​36570)
• Fix stranded row content under Node stream backpressure (@​unstubbable #​36516)
• Fix dropped FormData entries in decodeReplyFromBusboy (@​unstubbable #​36468)
• Harden types and improve performance of Flight replies (@​eps1lon #​36425)
• Add more cycle protections to Flight replies (@​eps1lon #​36236)
• Transport AggregateError.errors (@​eps1lon #​36156)
• Bail out early when a backing entry for Blob deserialization is not a Blob (@​eps1lon #​36055)
• Clear chunk reason after successful module initialization (@​unstubbable #​36024)
• Add support for transporting Error.cause (@​eps1lon #​35810)
• Walk parsed JSON instead of using a reviver for faster RSC payload parsing (@​timneutkens #​35776)
• Skip transferReferencedDebugInfo during debug info resolution (@​unstubbable #​35795)
• Add unstable_allowPartialStream option to Flight Client (@​unstubbable #​35731)
• Fix encodeReply for JSX with temporary references (@​unstubbable #​35730)
• Fix debug channel flag in Node.js renderToPipeableStream (@​feedthejim #​35724)
• Fix stack overflow in visitAsyncNode with deep async chains (@​unstubbable #​35612)
• Allow Server Components to render context providers from client modules (@​rickhanlonii #​35675)
• Allow overriding request.timeOrigin via options.startTime (@​lubieowoce #​35598)
• Warn once if eval is disabled in the dev environment (@​eps1lon #​35661)
• Restore original function name in dev server callstacks served with unsafe-eval (@​eps1lon #​35650)
• Fix react-markup types for server references (@​eps1lon #​35634)
• Add more DoS mitigations to Flight replies and harden Flight (@​unstubbable #​35632)
• Allow cyclic references to be serialized when unwrapping lazy elements (@​unstubbable #​35471)
• Improve cyclic thenable detection in Flight reply server (@​cvan20191 #​35369)
• Add extra loop protection to Flight reply server (@​sebmarkbage #​35351)
• Patch Promise cycles and toString on Server Functions (@​sebmarkbage #​35345)
• Move react-server-dom-webpack/*.unbundled to private react-server-dom-unbundled (@​eps1lon #​35290)
• Never parse then functions in Flight replies (@​sebmarkbage #​35289)
• Patch Flight reply server with fixes from Flight Client (@​sebmarkbage #​35277)
• Complete the list of Node.js internal Promise awaits (@​eps1lon #​35161)
• Reduce risk of maximum call stack exceeded when emitting async sequences (@​eps1lon #​35159)
• Fix pending chunks count for streams and async iterables in DEV (@​unstubbable #​35143)
• Fix broken byte stream parsing caused by buffer detachment (@​unstubbable #​35127)
• Fix hasReadable flag in Node.js clients' debug channel (@​unstubbable #​35039)
• Fix debug info filtering to include later resolved I/O (@​unstubbable #​35036)
• Cache the value when visiting the same I/O or Promise along different paths (@​sebmarkbage #​35005)
• Clone subsequent I/O nodes when resolved more than once (@​sebmarkbage #​35003)
• Fix hang after resolving cyclic references (@​unstubbable #​34988)
• Allow passing DEV-only startTime as an option (@​sebmarkbage #​34912)
• Ignore bound-anonymous-fn resources as they're not considered I/O (@​sebmarkbage #​34911)
• Fix detached ArrayBuffer error when streaming typed arrays (@​unstubbable #​34849)
• Exclude RSC stream if the stream resolves in a task (@​sebmarkbage #​34838)
• Forward the current environment when forwarding I/O entries (@​sebmarkbage #​34836)
• Look for moved debug info when logging component performance track (@​sebmarkbage #​34839)
• Respect displayName of Promise instances on the server (@​sebmarkbage #​34825)
• Avoid unnecessary indirection when serializing debug info (@​unstubbable #​34797)
• Fix preload as attribute for stylesheets (@​unstubbable #​34760)
• Don't serialize toJSON in debug path and omit wide arrays (@​sebmarkbage #​34759)
• Add support for <Activity> (@​eps1lon #​34697)

v19.2.8: 19.2.8 (July 21st, 2026)

Compare Source

React Server Components

• Performance improvements when decoding
  (#​37087 by @​eps1lon)

v19.2.7

Compare Source

React Server Components

• Fixed missing FormData entries in Server Actions which regressed in 19.2.6 (@​unstubbable #​36566)

v19.2.6

Compare Source

React Server Components

• Type hardening and performance improvements (@​eps1lon, @​unstubbable #​36425)

v19.2.5

Compare Source

React Server Components

• Add more cycle protections (@​eps1lon, @​unstubbable #​36236)

v19.2.4

Compare Source

React Server Components

• Add more DoS mitigations to Server Actions, and harden Server Components (@​gnoff, @​lubieowoce, @​sebmarkbage, @​unstubbable #​35632)

v19.2.3

Compare Source

React Server Components

• Add extra loop protection to React Server Functions (@​sebmarkbage #​35351)

v19.2.2

Compare Source

React Server Components

• Move react-server-dom-webpack/*.unbundled to private react-server-dom-unbundled (@​eps1lon #​35290)
• Patch Promise cycles and toString on Server Functions (@​sebmarkbage, @​unstubbable #​35289, #​35345)

v19.2.1

Compare Source

React Server Components

• Bring React Server Component fixes to Server Actions (@​sebmarkbage #​35277)

v19.2.0

Compare Source

Below is a list of all new features, APIs, and bug fixes.

Read the React 19.2 release post for more information.

New React Features

• <Activity>: A new API to hide and restore the UI and internal state of its children.
• useEffectEvent is a React Hook that lets you extract non-reactive logic into an Effect Event.
• cacheSignal (for RSCs) lets your know when the cache() lifetime is over.
• React Performance tracks appear on the Performance panel’s timeline in your browser developer tools

New React DOM Features

• Added resume APIs for partial pre-rendering with Web Streams:
  • resume: to resume a prerender to a stream.
  • resumeAndPrerender: to resume a prerender to HTML.
• Added resume APIs for partial pre-rendering with Node Streams:
  • resumeToPipeableStream: to resume a prerender to a stream.
  • resumeAndPrerenderToNodeStream: to resume a prerender to HTML.
• Updated prerender APIs to return a postponed state that can be passed to the resume APIs.

Notable changes

• React DOM now batches suspense boundary reveals, matching the behavior of client side rendering. This change is especially noticeable when animating the reveal of Suspense boundaries e.g. with the upcoming <ViewTransition> Component. React will batch as much reveals as possible before the first paint while trying to hit popular first-contentful paint metrics.
• Add Node Web Streams (prerender, renderToReadableStream) to server-side-rendering APIs for Node.js
• Use underscore instead of : IDs generated by useId

All Changes

React

• <Activity /> was developed over many years, starting before ClassComponent.setState (@​acdlite @​sebmarkbage and many others)
• Stringify context as "SomeContext" instead of "SomeContext.Provider" (@​kassens #​33507)
• Include stack of cause of React instrumentation errors with %o placeholder (@​eps1lon #​34198)
• Fix infinite useDeferredValue loop in popstate event (@​acdlite #​32821)
• Fix a bug when an initial value was passed to useDeferredValue (@​acdlite #​34376)
• Fix a crash when submitting forms with Client Actions (@​sebmarkbage #​33055)
• Hide/unhide the content of dehydrated suspense boundaries if they resuspend (@​sebmarkbage #​32900)
• Avoid stack overflow on wide trees during Hot Reload (@​sophiebits #​34145)
• Improve Owner and Component stacks in various places (@​sebmarkbage, @​eps1lon: #​33629, #​33724, #​32735, #​33723)
• Add cacheSignal (@​sebmarkbage #​33557)

React DOM

• Block on Suspensey Fonts during reveal of server-side-rendered content (@​sebmarkbage #​33342)
• Use underscore instead of : for IDs generated by useId (@​sebmarkbage, @​eps1lon: #​32001, facebook#33342#​33099, #​33422)
• Stop warning when ARIA 1.3 attributes are used (@​Abdul-Omira #​34264)
• Allow nonce to be used on hoistable styles (@​Andarist #​32461)
• Warn for using a React owned node as a Container if it also has text content (@​sebmarkbage #​32774)
• s/HTML/text for for error messages if text hydration mismatches (@​rickhanlonii #​32763)
• Fix a bug with React.use inside React.lazy-ed Component (@​hi-ogawa #​33941)
• Enable the progressiveChunkSize option for server-side-rendering APIs (@​sebmarkbage #​33027)
• Fix a bug with deeply nested Suspense inside Suspense fallback when server-side-rendering (@​gnoff #​33467)
• Avoid hanging when suspending after aborting while rendering (@​gnoff #​34192)
• Add Node Web Streams to server-side-rendering APIs for Node.js (@​sebmarkbage #​33475)

React Server Components

• Preload <img> and <link> using hints before they're rendered (@​sebmarkbage #​34604)
• Log error if production elements are rendered during development (@​eps1lon #​34189)
• Fix a bug when returning a Temporary reference (e.g. a Client Reference) from Server Functions (@​sebmarkbage #​34084, @​denk0403 #​33761)
• Pass line/column to filterStackFrame (@​eps1lon #​33707)
• Support Async Modules in Turbopack Server References (@​lubieowoce #​34531)
• Add support for .mjs file extension in Webpack (@​jennyscript #​33028)
• Fix a wrong missing key warning (@​unstubbable #​34350)
• Make console log resolve in predictable order (@​sebmarkbage #​33665)

React Reconciler

• createContainer and createHydrationContainer had their parameter order adjusted after on* handlers to account for upcoming experimental APIs

v19.1.9: 19.1.9 (July 21st, 2026)

Compare Source

React Server Components

• Performance improvements when decoding
  (#​37088 by @​eps1lon)

v19.1.8

Compare Source

React Server Components

• Fixed missing FormData entries in Server Actions which regressed in 19.1.7 (@​unstubbable #​36567)

[`v1

❗ Important

✂ PR body was truncated to here.

v1.1.1

Compare Source

• Fixed render performance (by @​Seigiard).

v1.105.1

Compare Source

• Improve error messages for @extends across different media queries.

v8.3.3

Compare Source

Bug Fixes

• deps: update launch-editor to v2.14.2 (#​23654) (22fd1d5)
• html: filename passed to transformIndexHtml should not include queries (#​23653) (7dafd8e)
• server: check fs.serve for ?vite-wasm-instance (#​23655) (ba8b7ab)
• server: store ids to safeModulePaths rather than URLs (#​23656) (c3e06f9)

v5.0.3

Compare Source

   🐞 Bug Fixes

• Isolate result.status between repeats runs  -  by @​hi-ogawa, Hiroshi Ogawa and Codex (GPT-6) in #​11218 (5dbeb)
• Don't print an interceptor warning in browser mode  -  by @​sheremet-va in #​11377 (15cc0)
• Don't retry when test.fails expectedly failed  -  by @​hi-ogawa, Hiroshi Ogawa and Codex (GPT-6) in #​11219 (b2458)
• Scope cache key generators to projects  -  by @​ecoyoung in #​11281 and #​11301 (92ba7)
• browser:
  • Delay server listen until tests start running  -  by @​sheremet-va in #​11366 (7d8ed)
  • Check mock path boundaries  -  by @​saryn17, Ryosei Sato and @​sheremet-va in #​11361 and #​11362 (1c388)
  • Keep config of browser-consumed environments  -  by @​kasperpeulen and Claude in #​11378 (aafc0)
  • Ignore page crash while cancelling  -  by @​sheremet-va in #​11386 (7c367)
  • toMatchScreenshot uses wrong reference on retried tests  -  by @​macarie in #​11393 (c22ab)
• cache:
  • Revalidate imports of cached modules  -  by @​sheremet-va in #​11381 (38f98)
• deps:
  • Pin why-is-node-running to 3.2.1 to avoid users running into ERR_PNPM_TRUST_DOWNGRADE  -  by @​AriPerkkio in #​11403 (f6c9a)
• expect:
  • Pass current equality testers to expect.extend asymmetric matchers  -  by @​hi-ogawa, Hiroshi Ogawa and Claude in #​11401 (3e794)
• jsdom:
  • Support Blob on jsdom 30.1  -  by @​sheremet-va in #​11379 (6c49b)
• pool:
  • Preserve unique pool ids when groupOrder is set  -  by @​mtorp in #​11392 (50312)
• ui:
  • Split-pane handle overlapping iframe  -  by @​macarie in #​11221 (f91db)
• vitest:
  • Remove root temp dir on close  -  by @​abhinav-phi in #​11248 (7c711)
• vm:
  • Do not optimize deps from index.html  -  by @​ezefernandezyf, Hiroshi Ogawa and Codex (GPT-6) in #​11329 and #​11360 (caf28)
  • Don't reuse scripts across vite environments  -  by @​MO2k4, Martin Oehlert and Claude in #​11395 (346d3)

    View changes on GitHub

v12.0.3

Compare Source

🩹 Fixes

• keep watching when the last test file is removed (#​6355) (921c161)
• show require() error on unsupported directory import (#​6354) (a68344f)
• support --X one-char aliases (#​6391) (1227939)

📚 Documentation

• refresh help output, fix stale links (#​6357) (bb2e69a)

🧹 Chores

• add --import=tsx integration test (#​6356) (3382fdd)
• touch a new file without stamping it afterwards (#​6349) (d59467a)

v18.2.0

Compare Source

Features

• support creating completions for fish shell (#​2568) (34d833a)

Bug Fixes

• completion: remove spurious single-quote in zsh loadautofunc detection (#​2548) (224e285)

v6.1.2

Compare Source

Fix HMR for compound components (#​1484)

Adding or removing a component from a compound component was not handled, creating stale updates. Doing this now triggers hmr invalidate on the file and propagates the invalidation to importers.

Fix compiler.logDiagnostics option not working (#​1483)

With oxc-transform-react >= 0.148, the logDiagnostics was not working. This version fixes it and requires oxc-transform-react >= 0.152 as a peer dependency if you are using the compiler option. compiler.logDiagnostics option is now deprecated and should be swapped for the builtin compiler.reportDiagnostics.

Disable refresh for non-jsx with compiler enabled (fix #​1478) (#​1485)

When the compiler was enabled, TS files could trigger fast refresh which would cause runtime errors. The logic now follows what the builtin oxc plugin in Vite does.

v0.152.0

🚀 Features

• 3255b89 transform-react: Add reportDiagnostics option to expose recoverable React Compiler diagnostics (#​26624) (Jack)
• f585cb4 napi: Add threadless WASI builds (#​26898) (Boshen)

v2.11.3

Compare Source

Patch Changes

• #​991 04445ba Thanks @​JoviDeCroock! - Don't re-run a useSignalEffect for a component that was unmounted by the same update. Preact 11 disposes it after paint, so the effect could otherwise run once more against an unmounted component.

v6.8.0

Compare Source

Minor Changes

• #​472 2dff53d Thanks @​jacob-ebey! - Replaces the second context argument in our streaming functions with an options object, the context moves to the third position.

  Streaming is considered alpha so we're making this breaking change deliberately as part of a minor.

Patch Changes

• #​464 455daf5 Thanks @​JoviDeCroock! - Call options._root before the render-to-string invocation

• #​475 4e92059 Thanks @​JoviDeCroock! - Use Node's Writable type for renderToPipeableStream destinations.

• #​444 e5ff19f Thanks @​dsherret! - fix: use cjs types instead of esm for jsx export to fix default export from esm

• #​477 4b80672 Thanks @​JoviDeCroock! - Preserve client-rendered content when delayed streaming chunks arrive after hydration.

• #​471 0ce2b84 Thanks @​JoviDeCroock! - Keep Preact option hooks isolated for the lifetime of asynchronous renders.

• #​476 5b1d509 Thanks @​JoviDeCroock! - Replace streamed Suspense fallbacks when the resolved content is empty.

v1.62.5

Compare Source

Patch Changes

• Updated dependencies [7f700ef, 90e6a1b, 7f57b1c, 6a4b0fe]:
  • wrangler@​4.147.0
  • miniflare@5.20261001.0-alpha

v1.62.4

Compare Source

Patch Changes

• Updated dependencies [b8e7cc3, b00ef4f, 9d7b08e, 464a582, aee2842, efd67e6]:
  • wrangler@​4.146.0
  • miniflare@5.20261001.0-alpha

v1.62.3

Compare Source

Patch Changes

• Updated dependencies [b9f1cdc, ddaa558, 8468487, a0712e5, a0712e5]:
  • miniflare@5.20260930.0-alpha
  • wrangler@​4.145.0

v1.62.2

Compare Source

Patch Changes

• Updated dependencies [91a3606, 2a15ae2]:
  • wrangler@​4.144.0
  • miniflare@5.20260926.1-alpha

v1.62.1

Compare Source

Patch Changes

• #​15938 62fd03a Thanks @​dieub! - Resolve the affected Undici dependency in new Wrangler and Vite plugin installs

  Undici 7.29.1 fixes GHSA-3wwx-pv8p-q78v. Update the shared dependency catalog and matching types used by Miniflare and Wrangler so downstream installs can resolve the patched runtime without an application-level override. A published release is still required for consumers; this changeset does not alter already published package metadata.

• Updated dependencies [7bb6eae, 60ccdbd, 62fd03a, 06ed9c8, 86211fe, c2bb4c8, eb1efe0, 485cfb3]:

  • wrangler@​4.143.1
  • miniflare@5.20260926.1-alpha

v1.62.0

Compare Source

Minor Changes

• #​15914 7f0734c Thanks @​jamesopstad! - Use cf/config for cloudflare.config.ts authoring

  Experimental cloudflare.config.ts projects must now import defineConfig, bindings, triggers, and related helpers from cf/config. Generated declarations from Wrangler and the Vite plugin also reference this package, so projects using the experimental configuration flow must add cf as a dependency.

  The Vite plugin no longer exports @cloudflare/vite-plugin/experimental-config. wrangler/experimental-config remains available for defineWranglerConfig, but no longer re-exports Cloudflare configuration helpers.

Patch Changes

• #​15878 e7915c1 Thanks @​dawNotPoi! - Keep dependency optimization caches stable on the first Vite dev server restart.

  The first dev server restart no longer re-optimizes unchanged dependencies, including in projects without Containers. Container images are still cleaned up when the server closes, even after a config reload removes the Cloudflare plugin.

• Updated dependencies [7f0734c]:

  • wrangler@​4.143.0

v1.61.0

Compare Source

Minor Changes

• #​15856 4c2993b Thanks @​Naapperas! - Support Workflows declared in exports on ctx.exports in local development

  A Workflow declared in a Worker's exports is now available on ctx.exports in wrangler dev, the Vite plugin and the Vitest plugin, with the same API as a Workflow binding:

  const instance = await ctx.exports.MyWorkflow.create({
    params: { name: "World" },
  });

  ctx.exports and workflows bindings with the same Workflow name share their instances, including instances created before the Workflow was declared in exports. Two Workers can't export the same Workflow name, and a binding to an exported Workflow must refer to the Worker and class that export it. getPlatformProxy() ignores Workflows declared in exports, since it doesn't run the Worker's code.

  wrangler workflows commands run with --local also work with Workflows declared only in exports, without a workflows binding.

  In the Vitest plugin, introspectWorkflow() and introspectWorkflowInstance() still need a Workflow binding, and now explain how to add one when passed a Workflow from ctx.exports. Instances created through ctx.exports are introspected too. A workflows binding whose script_name is the Worker's own name now resolves to the Worker itself again.

Patch Changes

• Updated dependencies [8dc53ae, 4c2993b]:
  • miniflare@5.20260926.0-alpha
  • wrangler@​4.142.0

v1.60.2

Compare Source

Patch Changes

• Updated dependencies [ee2b200, c91279b, 8280086, bd56b98, 805af2f]:
  • miniflare@5.20260925.0-alpha
  • wrangler@​4.141.0

v1.60.1

Compare Source

Patch Changes

• Updated dependencies [8f7916c]:
  • wrangler@​4.140.0
  • miniflare@5.20260923.0-alpha

v1.60.0

Compare Source

Minor Changes

• #​15648 52c0e9f Thanks @​tpmmorris! - Advertise Local Explorer scheduled invocations to headless agents

  The Vite plugin's Local Explorer hint now documents how to invoke a Worker's scheduled handler and points agents to the OpenAPI schema for the complete request contract.

Patch Changes

• Updated dependencies [52c0e9f, 44f5295, be72815, 479e1e8, 940c692, 52c0e9f, cd60c9c, 15799d4, bdda4c3, fc3cbaa]:
  • miniflare@5.20260923.0-alpha
  • wrangler@​4.139.0

v1.59.0

Compare Source

Minor Changes

• #​15817 6e77c53 Thanks @​jamesopstad! - Allow framework commands to produce Preview Build Output with the experimental config

  When cf previews deploy invokes a framework build command, Preview intent is now preserved. Function-based cloudflare.config.ts files receive isPreview: true, and generated Build Output is marked as a Preview build.

Patch Changes

• #​15806 8fade73 Thanks @​NuroDev! - Standardize Zod validation error output

  Format validation errors with Zod's built-in prettifyError() helper so Miniflare, Wrangler, the Vite plugin, and the Vitest plugin show consistent messages and property paths.

• Updated dependencies [a71237a, b03f960, 8fade73, 6e77c53]:

  • miniflare@5.20260921.1-alpha
  • wrangler@​4.138.0

v1.58.0

Compare Source

Minor Changes

• #​15778 cd7508c Thanks @​jamesopstad! - Generate types during development and supported builds with Vite's experimental.newConfig option or Wrangler's --experimental-new-config flag (and --experimental-cf-build-output for builds)

  When Wrangler's --experimental-new-config flag or Vite's experimental.newConfig option is enabled, inferred configuration and runtime declarations are now kept in .cloudflare/types/index.d.ts. Vite refreshes them during development and production builds. Wrangler refreshes them during development and when building with both --experimental-new-config and --experimental-cf-build-output. In the experimental wrangler.config.ts format, the types option is now top-level because it applies to both commands.

Patch Changes

• #​15765 1bdb96d Thanks @​th0m! - Prepare the required egress sidecar for local Containers without configured images

  Wrangler dev and Vite dev/preview now pull the required sidecar for Durable Object-managed Containers that select their application image at start time. Previously, these Containers failed to start unless the sidecar image was already cached in Docker.

• Updated dependencies [1bdb96d, cd7508c, f5605f5]:

  • wrangler@​4.137.0

v1.57.3

Compare Source

Patch Changes

• Updated dependencies [59267fc, 6906bf0, 354ebdb]:
  • wrangler@​4.136.3
  • miniflare@5.20260921.0-alpha

v1.57.2

Compare Source

Patch Changes

• Updated dependencies [ad20547, 02c1d83, 275184d, bd59eca]:
  • wrangler@​4.136.2
  • miniflare@5.20260921.0-alpha

v1.57.1

Compare Source

Patch Changes

• Updated dependencies [0ed4c54, 703922d, 14d946d]:
  • wrangler@​4.136.1
  • @​cloudflare/unenv-preset@2.16.2

v1.57.0

Compare Source

Minor Changes

• #​15713 3c75cad Thanks @​jamesopstad! - Identify experimental Build Output resource configs by filename and location

  The root remains config.json, Worker configs are now worker.config.json, and Container configs are now container.config.json. Resource configs no longer contain top-level type discriminators, while settings and build context are stored together in the root config.

• #​15713 3c75cad Thanks @​jamesopstad! - Define experimental Cloudflare configuration with a single default export

  Experimental cloudflare.config.ts files now define settings and resources together in a default-exported defineConfig() call. Add a Worker under worker, add Containers to the containers array, or omit both to provide settings only.

  import * as entrypoint from "./src/index.ts" with { type: "cf-worker" };
  
  export default defineConfig({
  	accountId: "...",
  	complianceRegion: "public",
  	worker: {
  		name: "my-worker",
  		compatibilityDate: "2026-09-18",
  		entrypoint,
  	},
  });

Patch Changes

• #​15711 91e2f86 Thanks @​ghostwriternr! - Allow local Container images without exposed ports

  Wrangler and the Cloudflare Vite plugin no longer reject images that omit Docker EXPOSE metadata. Local Containers can run command-only workloads or serve traffic through workerd without declaring an unused image port.

• Updated dependencies [3c75cad, 3c75cad, 91e2f86, c5913a6, 35668d7, 3c75cad, 45b3b81, 8235e6a, 0751490, 43b1f85, 731a2ee]:

  • wrangler@​4.136.0
  • miniflare@5.20260921.0-alpha

v1.56.0

Compare Source

Minor Changes

• #​15609 1f070c8 Thanks @​emily-shen! - Build Containers when emitting experimental Build Output

  Wrangler and the Cloudflare Vite plugin now build Dockerfile-backed Container images when experimental Build Output is enabled. Container configs are emitted under .cloudflare/output/v0/containers with local image references, while existing registry references pass through unchanged.

• #​15329 c4c9b75 Thanks @​akshitsinha! - Evaluate Flagship flags locally during development

  Flagship bindings now use the local Miniflare store by default in Wrangler and the Vite plugin, keeping development offline and isolated from production flags. Set remote: true on a binding to continue using its remote app.

  Use wrangler flagship flags pull <APP_ID> to seed the store from a remote app. Flag management commands also accept --local to read and update the local store directly.

• #​15701 643e5cc Thanks @​WillTaylorDev! - Pass Preview intent to defineWorker and upload its resolved configuration

  Preview builds now evaluate programmatic Worker configuration with ctx.isPreview set to true and record that intent in Build Output. The shared Preview uploader deploys the resolved bindings and settings while preserving configured Preview base values when it creates a Preview.

Patch Changes

• Updated dependencies [1f070c8, a0485d5, 629ddef, c4c9b75, 643e5cc]:
  • wrangler@​4.135.0
  • miniflare@5.20260918.0-alpha

v1.55.0

Compare Source

Minor Changes

• #​15674 a939a6c Thanks @​ghostwriternr! - Support explicit named Container image selection in local development

  The Vite plugin builds or pulls named images from Wrangler configuration and exposes their local tags through ctx.container.images. Pass one of those references to ctx.container.start({ image }) to select the image.

  This extends the experimental Durable Object-managed Containers interface.

Patch Changes

• Updated dependencies [6874aa9, 2298cf1, d96b319, 876eea1, 2b39fc2]:
  • wrangler@​4.134.0
  • miniflare@5.20260917.0-alpha

v1.54.11

Compare Source

Patch Changes

• Updated dependencies [71b6f10, ad23e6e, bac0c6a, be2437a, bac0c6a, 6f3d7b5]:
  • miniflare@5.20260916.0-alpha
  • wrangler@​4.133.0

v1.54.10

Compare Source

Patch Changes

• #​14775 1be7b97 Thanks @​dario-piotrowicz! - Sync Local Explorer endpoint lists across agent hints

  The Local Explorer endpoint list is now consistent across the three places it appears: the AGENTS.md template in create-cloudflare, the runtime agent hint in wrangler dev, and the Vite plugin agent hint. All three now include the observability/clear endpoint, use the canonical /cdn-cgi/local/explorer path, and have cross-reference comments pointing to each other.

• #​15399 982b806 Thanks @​tpmmorris! - Improve over-limit run_worker_first errors when duplicate rules are present

  The error now reports distinct and duplicate-entry counts and lists duplicated rules, making it clear when removing redundant entries can bring the configuration within the limit.

  Too many `run_worker_first` rules were provided; 105 rules provided (99 distinct, 6 duplicate entries) exceeds max of 100. Note: duplicate entries count towards the route limit. Ensure that no duplicate rules are present in your `run_worker_first` configuration.
  
  The duplicated rules found are:
  - "/rule/0"
  - "/rule/1"
  - "/rule/2"
  - "/rule/3"
  - "/rule/4"
  ...and 1 more duplicated rule.
  

• Updated dependencies [1be7b97, b149147, 7db596c, 76c0ce6, a83d7ac, a0856da, e35c4a1, d3565a5, a83d7ac, cb0955f, fa79b26, 16d1310, ca71205, 1015cfb, 982b806, ffabe74, 7db596c, e03822a, a83d7ac, 641df47, c4a6279]:

  • wrangler@​4.132.0
  • miniflare@5.20260915.0-alpha

v1.54.9

Compare Source

Patch Changes

• #​15574 164e4fb Thanks @​RealBhupesh! - Ignore .wrangler persistence writes in Vite's file watcher

  Miniflare stores local D1, KV, R2, and observability state under .wrangler/state. Those writes were watched as source changes on Linux and Windows, which fired every plugin hotUpdate hook and could make page loads take seconds. The plugin now ignores **/.wrangler/** while preserving any server.watch.ignored patterns already set by the user.

• Updated dependencies [8997652]:

  • miniflare@5.20260911.1-alpha
  • wrangler@​4.131.2

v1.54.8

Compare Source

Patch Changes

• Updated dependencies [945aaa3, 945aaa3, 47d906f, c2699bf, 945aaa3, 945aaa3]:
  • wrangler@​4.131.1
  • miniflare@5.20260911.0-alpha

v1.54.7

Compare Source

Patch Changes

• #​15432 f45b596 Thanks @​razethion! - Prevent delayed internal errors from fetch-only remote bindings

  Fetch-only remote bindings such as D1 and R2 previously opened an unused WebSocket RPC session. RPC sessions are now created only when an RPC method is called.

• Updated dependencies [0b43395, 9d75006, b605aa6, f45b596, f69f95a, a549e58, 36aed7f, 493e635, bff525d, dbb3ff4, fea3cd0, 96688b3, 6bd7b6c, 15cd6e1, be1caec, ed5797a, dbc9506, 128235a, f8aea7e, 24ef86b]:

  • wrangler@​4.131.0
  • miniflare@5.20260910.0-alpha

v1.54.6

Compare Source

Patch Changes

• Updated dependencies [edb3631, dbf6aad, bcebf08, e20df20, e20df20, 63c7ff1, a4e41df]:
  • miniflare@5.20260908.0-alpha
  • wrangler@​4.130.0

v1.54.5

Compare Source

Patch Changes

• #​15519 e004845 Thanks @​devaniketh! - Preserve HTTP/2 :authority header and non-default port in dev server requests

  When Vite runs over HTTPS with HTTP/2 enabled, browsers send authority via the :authority pseudo-header rather than Host. Previously, pseudo-headers were omitted when creating Fetch requests, causing non-default ports to be dropped from request.url and X-Forwarded-Host. Authority and scheme are now preserved from HTTP/2 pseudo-headers and request properties.

• Updated dependencies [8bbcb9f, 2b42d6f, ea5634e, c0c6504, ffc7efd, 682cd44]:

  • miniflare@5.20260907.0-alpha
  • wrangler@​4.129.1

v1.54.4

Compare Source

Patch Changes

• #​15407 7662118 Thanks @​xgame92! - Surface Worker export type fetch errors during development

  The Vite plugin now reports the Worker name, HTTP status, and response body when fetching export types fails. This preserves the underlying error instead of replacing it with a JSON parsing error.

• Updated dependencies [00a9f2f, 1dba24a, d40a634, 7c1b2a6, 93d72a5, f6fb347, f6fb347, dc24057, d2d8eea]:

  • miniflare@5.20260903.0-alpha
  • wrangler@​4.129.0

v1.54.3

Compare Source

Patch Changes

• Updated dependencies [dbbb795, 200780f, b3f2628, 87a7acf, 1809c5e, ea28cc3, b3fb2bf, 707cb6f]:
  • wrangler@​4.128.0
  • miniflare@5.20260831.0-alpha

v1.54.2

Compare Source

Patch Changes

• Updated dependencies [eb01850, e1df91a, b23de74, 015550a, 015550a, 015550a, 3650d29, b23de74]:
  • miniflare@5.20260828.0-alpha
  • wrangler@​4.127.1

v1.54.1

Compare Source

Patch Changes

• Updated dependencies [71b6f10, ad23e6e, bac0c6a, be2437a, bac0c6a, 6f3d7b5]:
  • miniflare@5.20260916.0-alpha
  • wrangler@​4.133.0

v1.54.0

Compare Source

Minor Changes

• #​15134 c66d2d5 Thanks @​gpanders! - Enable FUSE-capable local container development

  Miniflare now automatically passes the Docker privileges needed for FUSE to loc

❗ Important

✂ PR body was truncated to here.

v8.19.0

Compare Source

v3.0.3

Compare Source

What's Changed

• fix: handle TypeScript generic commas in exported bindings by @​BridgeAR in #​254
• fix: ASI and repeated non-null assertions in TypeScript export bindings by @​guybedford in #​257

Full Changelog: guybedford/es-module-lexer@3.0.2...3.0.3

v3.0.3

Compare Source

Patch Changes

• #​2297 3f163da Thanks @​Andarist! - Fixed semver ranges (such as >=1.0.0 <2.0.0) getting cut off (>=2.0.0) when updating internal dependencies.

• #​2276 ca9d110 Thanks @​Andarist! - Fixed pnpm 10 compatibility with npm 12 when reading registry information, packing, and publishing packages.

• Updated dependencies [3f163da, bfe9050, e522996]:

  • @​changesets/apply-release-plan@8.1.1
  • @​changesets/config@4.0.1

v5.7.1

Compare Source

What's Changed

• fix: emit perf maps when not parsing the full inlining logs from V8 by @​GuillaumeLagrange in #​85

Full Changelog: CodSpeedHQ/codspeed-node@v5.7.0...v5.7.1

v5.7.0

Compare Source

Highlights

• Dump full inlining information for higher optimization tiers in walltime so we can show more functions on the profiler. To test it out, set CODSPEED_WALLTIME_PROFILER=samply env variable in the codspeed action.
• Added support for tinybench v5 and v6 in @​codspeed/tinybench-plugin

What's Changed

• Add necessary flags to dump optimized inlining info and fix walltime marker placement by @​GuillaumeLagrange in #​83

Full Changelog: CodSpeedHQ/codspeed-node@v5.6.0...v5.7.0

v5.6.0

Compare Source

What's Changed

• feat: add MacOS support by @​not-matthias in #​81

Full Changelog: CodSpeedHQ/codspeed-node@v5.5.0...v5.6.0

v5.5.0

Compare Source

Highlights

We are introducing @codspeed/playwright, for walltime benchmarking and profiling of end to end browser applications through playwright.

Here's an example usage, but head to the docs for more information

import { bench, type Page } from "@codspeed/playwright-plugin";
import electronExecutable from "electron";
import path from "node:path";
import { fileURLToPath } from "node:url";

const __dirname = path.dirname(fileURLToPath(import.meta.url));
const appRoot = path.resolve(__dirname, "..");

async function waitUntilSettled(page: Page): Promise<void> {
  await page.waitForFunction(() => {
    const main = document.getElementById("main");
    return !!main && !main.classList.contains("loading");
  });
}

await bench(
  "inbox-search-archive-threads",
  async ({ page }) => {
    await page.fill("#search", "update");
    await waitUntilSettled(page);

    await page.click("#select-visible-btn");
    await page.click("#archive-btn");
    await waitUntilSettled(page);

    await page.click('#sidebar nav button[data-view="threads"]');
    await waitUntilSettled(page);
  },
  {
    target: {
      kind: "electron",
      appPath: path.join(appRoot, "out/main/index.js"),
      cwd: appRoot,
    },
    beforeRound: async ({ page }) => {
      page.setDefaultTimeout(180_000);
      await page.waitForSelector("#main");
      await waitUntilSettled(page);
    },
  },
);

Note: this plugin is only compatible with the walltime instrument.

What's Changed

• Add playwright profiling package by @​GuillaumeLagrange in #​80

Full Changelog: CodSpeedHQ/codspeed-node@v5.4.0...v5.5.0

v5.4.0

Compare Source

What's Changed

This release adds first support for macOS walltime.

Please note that profiling and other instruments are not yet available on macOS and will come in a later update.

• Support macos walltime without profiling in codspeed-node by @​GuillaumeLagrange in #​79

Full Changelog: CodSpeedHQ/codspeed-node@v5.3.0...v5.4.0

v5.3.0

Compare Source

What's Changed

We now collect buildtime and runtime environment data to warn users about differences in their runtime environment when comparing two runs against one another.

This data includes toolchain metadata like version and build options, as well as a list of dynamically loaded linked libraries.

• feat(fi): add memory profiling by @​not-matthias in #​73
• feat: dump node process information when running by @​GuillaumeLagrange in #​77
• chore: bump instrument-hooks to dump linked libraries by @​GuillaumeLagrange in #​78

Full Changelog: CodSpeedHQ/codspeed-node@v5.2.0...v5.3.0

v0.99.2

Compare Source

Added

• Added the lightweight @earendil-works/pi-ai/models entry point for model collections and provider construction without loading TypeBox, built-in catalogs, or provider SDKs.
• Added Anthropic workload identity federation from the Anthropic SDK environment variables ANTHROPIC_FEDERATION_RULE_ID, ANTHROPIC_ORGANIZATION_ID, and ANTHROPIC_IDENTITY_TOKEN_FILE (plus optional ANTHROPIC_SERVICE_ACCOUNT_ID and ANTHROPIC_WORKSPACE_ID). API keys and ANTHROPIC_AUTH_TOKEN take precedence (#​10177, #​10242 by @​philfreo)

Fixed

• Fixed context overflow detection for Z.AI CN endpoint Prompt exceeds max length errors (#​10208)
• Fixed Anthropic requests failing when a strict: "prefer" tool schema uses keywords Anthropic strict tool use rejects, such as minimum/maximum; such tools are now sent non-strict (#​9953)
• Fixed provider retries firing immediately when a Retry-After header contains an unparseable date; they now use exponential backoff (#​9571)

v0.99.1

Compare Source

Added

• Added GPT-6.1 Sol (gpt-6.1-sol) to the OpenAI, Azure OpenAI Responses, and OpenAI Codex providers.

v0.99.0

Compare Source

Breaking Changes

• Unified image models into the regular Provider/Models surface. The separate ImagesModels collection is removed: createImagesModels(), createImagesProvider(), ImagesProvider, openrouterImagesProvider(), builtinImagesProviders(), and builtinImagesModels() are gone. Use builtinModels(), models.getModelOfType("image", ...), models.generateImages(), and createProvider({ models, images }) instead. Existing unqualified reads remain chat-only.
• Image models are now ImageModel with a required type: "image" and share BaseModel with chat models. The old plural image type names (ImagesModel, ImagesApi, KnownImagesApi, KnownImagesProvider, and ImagesProviderId) are removed. generateImages() accepts only image models. Output modalities (output) remain on image models only.
• Generated model data schema is now version 6: every entry carries type, operation-specific catalogs include chat, image, and classifier models, and one upstream ID may have separate entries per type. OpenRouter image models live in the openrouter-images api group of openrouter.json; image-models.generated.ts and scripts/generate-image-models.ts are removed. Run npm run hydrate:model-data.

Added

• Added Models.generateImages() with provider-resolved auth, Provider.generateImages?, and createProvider({ images }) keyed by model.api. createProvider() models and fetchModels accept models of every type, and api is optional when images or classifiers is given.
• Added an optional model type ("chat", "image", or "classifier"). Chat models may omit it, so existing chat models, providers, and stores keep working unchanged. Narrow mixed lists with the new isModelType() guard or read the effective type with getModelType().
• Added getModelsOfType(), getModelOfType(), getAvailableOfType(), getAllModels(), and getAllAvailable() on Models; optional Provider.getAllModels() and Provider.filterAllModels(); corresponding generated-catalog accessors; and the AnyModel and ModelTypeMap types. hasApi(), calculateCost(), and modelsAreEqual() accept AnyModel.
• Added support for models of every type in ModelsStoreEntry.models. Stored and fetched models of unknown types are dropped instead of failing a refresh.
• Added classifier models and Models.classify() with a provider-neutral JEV-style choice/score/bool contract. The built-in TypeSafe provider exposes models.dev's jev-latest through the System One API and translates public bool questions to TypeSafe's noul wire format.
• Added Jev classifier models on OpenRouter (typesafe/jev-1.13, ~typesafe/jev-latest) through its TypeSafe-compatible System One endpoint, and on Cloudflare Workers AI (typesafe/jev) through the new cloudflare-workers-ai-system-one classifier API.
• Added Jev classifier models on Vercel AI Gateway (typesafe-ai/jev, generated from its evaluation model catalog) and OpenCode Zen (jev-1.13, jev-1.13-free) through their TypeSafe-compatible System One endpoints.
• Added usage to ClassifierResult: System One classifications report token counts, priced from the model catalog like chat usage.
• Added a runtime chat-model check to the Models stream entry points so non-chat models fail with a clear ModelsError instead of a missing-api stream error.
• Added array-based models.all.json and providers/{id}.all.json variants to the generated and published JSON catalog, allowing the same upstream ID once per model type; the existing keyed models.json and providers/{id}.json stay chat-only for released clients.
• Added onProviderStreamEvent to observe parsed provider stream events before normalization, including provider-specific fields not retained in assistant messages (#​9784, #​9901 by @​davidbrai).
• Added Claude Sonnet 5.5 to the built-in Anthropic model catalog with adaptive thinking, mid-conversation effort, 1M context, and official pricing metadata.
• Added Sign in with ChatGPT to the openai provider: an OAuth login that uses a ChatGPT subscription with the OpenAI API. Models.login() accepts LoginOptions with getDeviceId(), which supplies a stable installation ID to login flows that need one. Subscription usage-limit errors are not retried and link to the ChatGPT usage page; temporary usage errors are retried.
• Added the llama-cpp-classify classifier API, which answers classifier questions from llama-server's next-token probabilities for single-token answer labels.
• Added optional AssistantMessage.thinkingLevel, which records the thinking level the agent loop requested for a response.

Changed

• Renamed the OpenAI Codex provider to "OpenAI Codex (legacy)"; Sign in with ChatGPT on the openai provider supersedes it.
• Unified the Anthropic, OpenAI Codex, OpenRouter, and Radius browser sign-in callback servers into one shared implementation with the same browser pages. The OAuth page helpers are now available as @earendil-works/pi-ai/utils/oauth-page.
• Changed Radius browser sign-in to exchange the authorization code before showing the browser page, so token exchange failures are shown in the browser.

Fixed

• Fixed 1-hour Anthropic cache writes reported by Vercel AI Gateway in streaming deltas being priced at the 5-minute rate (#​9210).
• Fixed model-level samplingParams being dropped by direct stream()/complete() calls on OpenAI-compatible APIs (#​9506).
• Fixed Mistral GLM models producing empty text blocks and split thinking blocks from empty content deltas, which could make later requests fail with "Expected at most one leading ThinkChunk" (#​9674).
• Fixed OpenAI Fast mode requests being priced at the standard rate when the response reports service_tier: "fast", as GPT-6 models do (#​10034).
• Fixed Mistral reasoning models ignoring the requested thinking level: GLM 5.3 now uses reasoning_effort instead of prompt_mode, GLM 5.2 accepts max, and Mistral models only offer the effort levels the API supports (#​9678).
• Fixed OpenCode Zen and OpenCode Go qwen3.8-flash thinking being replayed as plain text on later turns because the endpoint returns empty thinking signatures (#​10047).
• Fixed OpenAI Responses streams returning unfinished tool calls as runnable, which made servers that omit output_index (such as llama.cpp) run mixed-up commands; such streams now end with an error (#​9974).
• Fixed Anthropic and OpenAI Codex browser sign-in waiting indefinitely after the provider redirected with an authorization error; sign-in now fails with the provider's error description.
• Fixed Anthropic browser sign-in failing when its callback port is in use; it now falls back to pasting the redirect URL.
• Fixed GitHub Copilot Claude Opus 5.5 offering unsupported thinking levels when upstream model metadata is incomplete; it now offers low through max.

v0.87.1

Compare Source

Added

• Added Claude Opus 5.5, GPT-6 Sol, and GPT-6 Luna to the GitHub Copilot catalog.
• Added GPT-6 Sol and GPT-6 Luna for OpenAI API keys and OpenAI Codex subscriptions, with full reasoning-effort, prompt-caching, tool-search, long-context pricing, and official cost metadata.
• Added Claude Opus 5.5 to the built-in Anthropic model catalog with adaptive thinking, 1M context, and official pricing metadata.
• Added Grok 4.7 to the built-in xAI model catalog with long-context pricing metadata.

Fixed

• Fixed image-only user messages being rejected by some OpenAI-compatible providers because they included an empty text part (#​9797)
• Fixed Anthropic OAuth requests reporting an outdated Claude Code version.

v0.87.0

Compare Source

Added

• Added model image-input limit and cache-safe resize metadata to the generated catalog (#​9631).

Fixed

• Fixed unknown OpenAI-compatible Chat Completions endpoints receiving strict tool schemas unless they explicitly advertise support, while preserving strict tools for capable built-in models (#​9816).

v0.86.1

Compare Source

Added

• Added Meta provider (Model API key and Muse subscription OAuth) with Muse Spark models (#​9096 by @​xl0).

Fixed

• Fixed z.ai Prompt too long errors not being recognized as context overflow (#​9805).
• Fixed Cerebras models advertising unsupported strict tool schemas, which caused HTTP 400 errors when strict and non-strict tools were mixed (#​9804 by @​EdenGottlieb).

v0.86.0

Compare Source

Breaking Changes

• Changed provider-facing ProviderStreams and StreamFunction inputs from Context to normalized TranscriptContext values. System prompts and tool declarations now live in transcript system messages; custom providers must read them with getCurrentSystemPrompt() and getCurrentTools(). See System Messages.
• Restricted ToolCall.arguments and ToolResultMessage.details to JSON-compatible values, changed ToolResultMessage into a conditional type, and made JsonValue arrays readonly.

Added

• Added transcript-backed mid-conversation system prompt and tool changes, with native replay on supported models and automatic collapse for other providers. See System Messages (#​9548).
• Added a generated public Radius model catalog for synchronous API lookup, with cached and live gateway catalogs overlaid at runtime.
• Enabled native deferred tool loading for Fireworks Messages models. Use ToolSearch or tool_search as the loader name for prompt-prefix deferral (#​9323).
• Added RetryPolicy.maxAgentDelayMs support to cap shared assistant retry backoff for summarization calls (#​8826).
• Added Model.promptCache lifetime metadata for short and long retention tiers to support prompt-cache warming decisions (#​9668).

Fixed

• Fixed GitHub Copilot GPT models, including GPT-6 Astra, using the Chat Completions adapter instead of the required Responses adapter (#​9253 by @​petrroll).
• Fixed DeepSeek V4.1 thinking levels on OpenRouter and OpenCode Go preserving provider effort metadata (#​9485).
• Fixed bodyless HTTP 400/413 errors from non-Cerebras providers being misclassified as context overflow (#​9482).
• Fixed Vercel AI Gateway replaying unsigned thinking as assistant text (#​9676).
• Fixed Google Generative AI and Vertex AI using unsupported thinking levels when reasoning is omitted or when model capabilities differ within a Gemini family (#​9455).
• Fixed Anthropic-compatible relays breaking signed thinking replay when they report a different response model, while preserving fallback pricing (#​9188).
• Fixed Amazon Bedrock one-hour cache writes being priced at the five-minute rate (#​9457).
• Fixed quadratic CPU usage when draining buffered EventStream events (#​9055).
• Fixed Mistral Medium reasoning requests to use reasoning_effort for all reasoning-capable mistral-medium-* model IDs instead of the unsupported prompt_mode (#​8700).
• Fixed OpenCode and OpenCode Go requests to send x-opencode-session from sessionId across all supported API adapters (#​9326).
• Fixed OpenAI Codex requests to send the model's Off reasoning effort instead of omitting it, while respecting unsupported Off mappings (#​9191).
• Fixed Fireworks unsigned thinking replay and reasoning effort selection using catalog metadata, with verified DeepSeek V4 and Qwen3.8 fallbacks and removal of redundant GLM 5.2 and Kimi K3 effort aliases (#​9323).
• Fixed OpenRouter requests to send x-session-id from sessionId for Chat Completions and Anthropic Messages models when prompt caching is enabled (#​9102).
• Fixed the DeepSeek catalog to advertise deepseek-flash for DeepSeek V4.1 Flash instead of retired Flash aliases, and refreshed DeepSeek pricing metadata (#​9423).
• Fixed Mistral-hosted GLM-5.2 reasoning requests to use reasoning_effort instead of the ignored prompt_mode (#​9375).
• Fixed OpenAI-compatible Responses errors to identify the actual provider instead of always labeling them as OpenAI errors (#​9298).
• Fixed Baseten requests to send session-affinity headers from sessionId for automatic prompt-cache routing (#​9629).
• Fixed retry classification for Cloudflare 520 responses (#​9627).
• Fixed retry classification for transient Azure peak-load capacity errors (#​9669).

Removed

• Removed GPT-5.4 and GPT-5.4 mini from the OpenAI Codex catalog after they became unavailable to ChatGPT accounts (#​9394).

v0.85.1

Compare Source

Added

• Added GPT-6 Astra for OpenAI API keys and OpenAI Codex subscriptions.

Fixed

• Fixed long prompt-cache requests for GPT-5.6+ Responses models to use prompt_cache_options.ttl: "30m" instead of prompt_cache_retention: "24h".

v0.85.0

Compare Source

Breaking Changes

• Replaced createGatewayBindingFetch() with createAiBindingFetch() for Cloudflare Workers AI bindings. Configure the model's Workers AI Gateway passthrough baseUrl directly; requests now pass through the binding unchanged (#​8287 by @​Maximo-Guk).

Added

• Added compact, persistable assistant-message frames with AssistantMessageFrameEncoder and reduceAssistantMessageFrames().
• Added the vllmPriority OpenAI-compatible model setting for forwarding scheduler priority to vLLM (#​9004 by @​AppleDannyClegg).
• Added the supportsMaxOutputTokens OpenAI Responses compatibility setting (#​8941 by @​scturtle).
• Added an optional timestamp argument to uuidv7() for follower IDs.
• Added narrow api, providers, and utils subpath exports for direct imports without loading the package barrel.
• Added Anthropic per-turn effort persistence, deterministic historical effort markers, and signed-thinking mismatch recovery for supported Claude models across Anthropic Messages transports, including OpenRouter.
• Added Meta provider (Model API key and Muse subscription OAuth) with Muse Spark models.

Fixed

• Removed the unavailable Grok Build 0.1 model from the built-in xAI catalog (#​9093 by @​Jaaneek).
• Fixed assistant-message frames preserving the provider thinking level.
• Fixed simple provider streams to consistently emit standard stream events and custom tool-call deltas.
• Fixed the Qwen Token Plan Individual catalog to include Qwen3.8 Flash (#​9021).
• Fixed Baseten GLM-5.2 models incorrectly advertising image input support (#​8293 by @​Panoplos).
• Fixed Fireworks GLM models using the wrong API adapter.
• Removed the unnecessary Chord dependency from pi-ai by defining its exported JsonValue type directly.
• Fixed GitHub Copilot Claude Fable 5 requests to use the Anthropic Messages adapter so selected reasoning levels are sent (#​8961).
• Fixed OpenAI Codex SSE parsing to process terminal events that are not followed by a blank line (#​9047).
• Fixed NO_PROXY matching for both root domains and subdomains (#​8737 by @​MeiSiristhebest).

v0.84.4

Compare Source

Added

• Added the experimental vision-capable deepseek-v4-flash-vision-exp model to the DeepSeek catalog.

Fixed

• Fixed OpenAI-compatible Chat Completions ignoring an explicitly requested toolChoice when no tools are defined.
• Fixed thinking signature serialization to run once after the signature is complete (#​8671).
• Fixed fragmented Mistral tool calls splitting when continuation chunks omit the tool-call ID (#​8387).
• Fixed OpenAI-compatible reasoning replay to merge consecutive streamed text and summary reasoning_details deltas.
• Fixed the Cloudflare AI Gateway catalog to include supported workers-ai/* passthrough models omitted by models.dev.
• Fixed OpenRouter reasoning controls by deriving off support and available effort levels from OpenRouter's model metadata, preventing reasoning-mandatory models from receiving effort: "none" (#​8614 by @​davidbrai).

v0.84.3

Compare Source

Breaking Changes

• Renamed GoogleThinkingLevel to GoogleApiThinkingLevel and added ResolvedGoogleThinkingLevel for normalized adapter levels.

Added

• Added provider-neutral toolChoice support to simple stream requests.
• Added automatic Anthropic server-side refusal fallback for supported first-party models, including returned-model usage pricing (#​8017).
• Added configurable OpenAI-compatible thinking-token budget fields for vLLM, Qwen/SGLang, and llama.cpp servers (#​8275 by @​bnsd55).
• Added China-specific ZAI Coding Plan models, including GLM-4.6V vision support, and API-equivalent usage cost estimates for models with published PAYG prices (#​8220).
• Added deepseek-v4-pro-0813 to the Qwen Token Plan Individual catalog (#​8194).

Changed

• Changed built-in xAI models to use the Responses API with encrypted reasoning replay and made Grok 4.6 the default xAI model (#​8124 by @​Jaaneek).
• Changed the Anthropic, Azure OpenAI, Google Generative AI, Google Vertex, Mistral, OpenAI Chat Completions, and OpenAI Responses adapters to send Pi's default User-Agent unless overridden (#​8305).

Fixed

• Fixed OpenAI-compatible Chat Completions reasoning replay to preserve and resend assistant-level reasoning_details (reasoning.text, reasoning.summary, and reasoning.encrypted) verbatim and in order (#​7994).
• Fixed Anthropic server-side fallback responses being priced with the requested model instead of the returned fallback model (#​8285).
• Fixed GitHub Copilot login triggering model-policy rate limits by limiting policy updates, retrying model discovery once, and honoring server retry delays (#​7850).
• Fixed Amazon Bedrock dropping and failing to replay opaque redacted reasoning from non-Anthropic models (#​8314 by @​seiji).
• Fixed Z.AI Coding Plan models deriving incomplete reasoning-effort metadata, including missing GLM-5.3 low, high, and max levels (#​8336).
• Fixed DeepSeek V4 Flash on OpenCode and OpenCode Go omitting its supported low thinking level (#​8181 by @​tianshuang).
• Fixed Azure OpenAI Responses ignoring toolChoice in provider-specific stream requests.
• Fixed Amazon Bedrock after_provider_response/onResponse to forward the raw response headers instead of only the synthesized request id header (#​8234).
• Fixed Kimi OpenAI-compatible usage reporting so top-level cached_tokens count as cache reads instead of normal input tokens (#​8075).
• Fixed Google Generative AI and Vertex AI custom models ignoring thinkingLevelMap, which dropped extended thinking controls (#​8135).
• Fixed Xiaomi model catalog generation retaining shut-down MiMo V2 model names after models.dev marked them deprecated (#​8187).
• Fixed OpenRouter reasoning controls by deriving off support and available effort levels from OpenRouter's model metadata, preventing reasoning-mandatory models from receiving effort: "none" (#​8454).

v0.84.2

Compare Source

Added

• Added createGatewayBindingFetch() for routing Cloudflare AI Gateway requests through a Workers AI binding without an API token (#​7901 by @​Maximo-Guk).
• Added AssistantMessage.endTurn to preserve OpenAI Codex's terminal end_turn signal for diagnostics (#​7766).

Changed

• Changed Kimi Coding requests to use pi's runtime User-Agent header.
• Automatically converted supported strict tool schemas to provider-compatible closed objects with required nullable optional fields while preserving original tool definitions, and treated null values for optional non-nullable tool arguments as omitted.
• Changed OpenAI Responses deferred tool loading to prefer message-anchored additional_tools where supported while retaining tool-search and top-level fallbacks (#​7709).
• Replaced the Mistral SDK transport with a native Chat Completions HTTP stream, eliminating its generated client and schema runtime overhead.

Fixed

• Fixed GitHub Copilot login triggering API rate limits while enabling model policies by limiting concurrent policy updates (#​6187).
• Fixed GitHub Copilot login still triggering API rate limits by updating only account models with unconfigured policies and honoring server retry delays (#​7850).
• Fixed upstream request buffer limit failures to trigger automatic assistant retries.
• Fixed OpenAI Responses function and custom tool calls to preserve namespaces during streaming, proxying, and replay (#​7709).
• Fixed built-in and custom DeepSeek API models to send output limits through the supported max_tokens field.
• Fixed Google Generative AI and Vertex AI responses with tool calls incorrectly treating output-limit or provider-error stops as normal tool use (#​8059).
• Fixed Amazon Bedrock replay rejecting tool arguments that contain empty object keys while preserving all valid nested values (#​7882 by @​muyiyr).
• Fixed DeepSeek compatibility detection for base URLs whose hostname contains uppercase letters (#​7933 by @​yearth).

v0.84.1

Compare Source

Added

• Added Qwen Token Plan Individual as a built-in provider with its documented subscription model catalog and the shared international QWEN_TOKEN_PLAN_API_KEY (#​7659 by @​arasovic).

v0.84.0

Compare Source

Breaking Changes

• Renamed the exported ModelsStreamTransforms interface to ModelsRequestTransforms because its header transformation now applies to all authenticated provider requests.

• Required dynamic model providers to accept a concrete RefreshModelsContext.signal; Models.refresh() remains unbounded when callers omit its optional signal.

• Required provider login, API-key check/resolution, and OAuth refresh implementations to accept a concrete abort signal; public auth and credential operations remain unbounded when callers omit their optional signal.

• Replaced raw RefreshModelsContext.store access with the read-only context.stored snapshot and generation-checked context.publish() transaction.

  createProvider({ fetchModels }): no catalog-publication migration is required. Before and after, return the fetched list; createProvider() restores stored models and publishes and persists refreshed models itself. signal is now guaranteed to be present.

  // Before
  const beforeProvider = createProvider({
    // ...
    fetchModels: async ({ signal }) => {
      const response = await fetch(catalogUrl, { signal });
      return parseModels(await response.json());
    },
  });
  
  // After: unchanged
  const afterProvider = createProvider({
    // ...
    fetchModels: async ({ signal }) => {
      const response = await fetch(catalogUrl, { signal });
      return parseModels(await response.json());
    },
  });

  Handwritten Provider.refreshModels(): replace direct store access and pre-publication mutation with generation-guarded publications.

  // Before
  refreshModels: async (context) => {
    const stored = await context.store.read();
    if (stored) currentModels = stored.models;
    if (!context.allowNetwork) return;
  
    const refreshed = await fetchModels(context.signal);
    currentModels = refreshed;
    await context.store.write({ models: refreshed, checkedAt: Date.now() });
  },
  
  // After
  refreshModels: async (context) => {
    if (context.stored) {
      const restored = context.stored.models;
      if (!(await context.publish({
        update: () => { currentModels = restored; },
      }))) return;
    }
    if (!context.allowNetwork) return;
  
    const refreshed = await fetchModels(context.signal);
    if (context.signal.aborted) return;
    await context.publish({
      persist: { models: refreshed, checkedAt: Date.now() },
      update: () => { currentModels = refreshed; },
    });
  },

  In publish(), omit persist to leave storage unchanged, pass a ModelsStoreEntry to write it, or pass persist: null to delete it. Omit update for metadata-only persistence; omit persist for an ephemeral in-memory publication.

Added

• Added optional OAuthAuth.isSubscription metadata for distinguishing subscription-backed authentication from generic OAuth sign-in.
• Added explicit TelemetryContext propagation across stream, deferred, and image request options using the vendor-neutral @earendil-works/pi-telemetry contract.
• Added deferred provider request contracts, durable response handles, authenticated fetch/cancel dispatch, and faux-provider support for pending, ready, failed, and cancelled responses (#​7339 by @​davidbrai).
• Added Baseten as a built-in OpenAI-compatible provider with models.dev catalog generation and native chat_template_args reasoning controls.
• Added arbitrary OpenAI-compatible sampling parameters through Model.samplingParams and StreamOptions.samplingParams, including per-request overrides (#​7568 by @​mrexodia).
• Added opt-in vLLM thinking_token_budget support for OpenAI-compatible models, reserving output tokens for the final answer (#​7638 by @​bnsd55).
• Added OpenAICompletionsCompat.supportsFinishReason for providers that omit streamed finish_reason values, inferring normal and tool-use stops when the stream ends.
• Added structured Amazon Bedrock failure diagnostics with HTTP status, modeled error code, and AWS request id when available (#​7286 by @​brianstanley).

Changed

• Added optional cancellation to ModelsStore reads, writes, and deletions; catalog orchestration binds these waits to the provider refresh signal.

Fixed

• Fixed GitHub Copilot Grok 4.5 requests to use the supported Responses API (#​7560).
• Bounded OAuth token refreshes so stalled requests release the credential-store lock (#​7508).
• Fixed tool argument validation to preserve values that already match an anyOf/oneOf union arm before attempting coercion, avoiding nullable unions converting null to another primitive value (#​7328).
• Fixed cancellation of model catalog refreshes so callers stop waiting even when a custom provider ignores its abort signal (#​7027).
• Fixed auth resolution, availability checks, OAuth refreshes, provider login, and in-memory credential queue waits to honor caller cancellation.
• Fixed newer provider refreshes being blocked by or overwritten by an older stalled generation, including persisted catalog publication.
• Updated GPT-5.6 Terra and Luna pricing across OpenAI and passthrough model catalogs.
• Fixed Fireworks Kimi K3 models to use the OpenAI-compatible API with native reasoning-effort levels and deferred tools (#​7199, #​7230 by @​XBeg9).
• Fixed Fireworks GLM 5.2 models sending the unsupported prompt_cache_retention field when long cache retention is enabled, and enabled session affinity for automatic prompt caching (#​7676).
• Updated Groq's Qwen reasoning override for the replacement qwen/qwen3.6-27b model.
• Fixed the OpenCode Go provider display name.
• Fixed provider error normalization treating arrays and class instances as structured response bodies instead of preserving their original errors (#​7205 by @​erikogenvik).
• Fixed Anthropic streams dropping text or thinking included in the initial content-block event (#​7358 by @​davidbrai).
• Fixed Google history conversion dropping signed empty text and thinking blocks required for replay (#​7362 by @​jingtao-wisdomgraph).
• Fixed OpenAI Codex cached WebSocket sessions being shared across different account credentials (#​7364).
• Fixed transient Google Generative AI and Vertex AI provider errors bypassing automatic retries (#​7471 by @​vish-pr).
• Fixed Gemini 3 tool call ids being discarded during history conversion, breaking signed multi-turn replay (#​7494 by @​muyiyr).
• Fixed OpenAI Responses incomplete reasons so only max_output_tokens is treated as a length stop, and exposed bounded recovery detection for responses truncated below their intended output limit (#​7540 by @​davidbrai).
• Restored GitHub Copilot models returned through account-specific policy responses (#​7672 by @​muyiyr).
• Replaced the retired Qwen Token Plan qwen3.8-max-preview model with qwen3.8-max (#​7670 by @​QuintinShaw).

v2.2.2

Compare Source

Patch Changes

• 2cb57be: Fix Claude models on the cloudflare-ai-gateway provider failing with 404 not_found_error: model: claude-sonnet-4.6 was not found. Since Pi 0.84.3 the gateway catalog lists Claude models with dotted versions (claude-sonnet-4.6), but the gateway's native Anthropic endpoint forwards the model ID to Anthropic, which only accepts dashed IDs. Flue now registers these models under Anthropic's IDs (cloudflare-ai-gateway/claude-sonnet-4-6); dotted specifiers still resolve to them. The cloudflare binding provider, which uses Cloudflare's dotted catalog names, is unchanged.

v2.2.1

Compare Source

Patch Changes

• 548937e: Fix model specifiers that stopped resolving after the Pi 0.87.1 up

❗ Important

✂ PR body was truncated to here.

v5.11.2

Compare Source

What's Changed

• fix: decode Uint8Array XML input as UTF-8 by @​emme1t in #​875
• fix: type EntityDecoderOptions.setXmlVersion's version as a number by @​YevheniiKotyrlo in #​880

New Contributors

• @​emme1t made their first contribution in #​875
• @​YevheniiKotyrlo made their first contribution in #​880

Full Changelog: NaturalIntelligence/fast-xml-parser@v5.11.1...v5.11.2

v5.11.1

Compare Source

What's Changed

• Bump actions/checkout from 7.0.0 to 7.0.1 by @​dependabot[bot] in #​858
• Bump zizmorcore/zizmor-action from 0.5.7 to 0.6.1 by @​dependabot[bot] in #​856
• Bump actions/setup-node from 6.4.0 to 7.0.0 by @​dependabot[bot] in #​854
• docs: fix two 404 documentation links by @​rajanpanth in #​862

New Contributors

• @​rajanpanth made their first contribution in #​862

Full Changelog: NaturalIntelligence/fast-xml-parser@v5.11.0...v5.11.1

v5.11.0

Compare Source

What's Changed

• add support for endIndex in node metadata (5.x edition) by @​Wain-PC in #​850
• fix: don't crash on a closing tag with no matching opening tag by @​hdimer in #​861

New Contributors

• @​Wain-PC made their first contribution in #​850
• @​hdimer made their first contribution in #​861

Full Changelog: NaturalIntelligence/fast-xml-parser@v5.10.1...v5.11.0

v5.10.1

Compare Source

Full Changelog: NaturalIntelligence/fast-xml-parser@v5.10.0...v5.10.1

v5.10.0

Compare Source

What's Changed

• Bump actions/checkout from 6.0.3 to 7.0.0 by @​dependabot[bot] in #​849
• Bump zizmorcore/zizmor-action from 0.5.6 to 0.5.7 by @​dependabot[bot] in #​848

Full Changelog: NaturalIntelligence/fast-xml-parser@v5.9.3...v5.10.0

v5.9.3

Compare Source

What's Changed

• Harden GitHub Actions workflows by @​martincostello in #​841
• GitHub Actions workflow fixes by @​martincostello in #​844

New Contributors

• @​martincostello made their first contribution in #​841

Full Changelog: NaturalIntelligence/fast-xml-parser@v5.9.2...v5.9.3

v5.9.2

Compare Source

Full Changelog: NaturalIntelligence/fast-xml-parser@v5.9.1...v5.9.2

v5.9.1

Compare Source

Full Changelog: NaturalIntelligence/fast-xml-parser@v5.9.0...v5.9.1

v5.9.0: update strnum, use is-unsafe

Compare Source

• update strnum to 2.3.0
  • you can set hex, binary, enotation, infinity, unicode
• validate unsafe HTML or XML data in doctype entities unsing 'is-unsafe' library. User can override rules by overriding EntityDecoder.

v1.9.0

Compare Source

   🚀 Features

• Add ignoreWorkspaceRootCheck option for pnpm/yarn v1  -  by @​danielroe in #​83 (8bb98)
• Why command  -  by @​himynameisdave and @​antfu in #​79 (d52fb)
• Add support for upm package manager  -  by @​iiio2, Claude Opus 5.5 and @​antfu in #​80 (bff26)
• Prefix bare package names with npm: for deno add  -  by @​antfubot in #​84 (55b09)
• Expose agent construction logic  -  by @​josh-hemphill and @​antfubot in #​65 (03a5e)
• Add minimumReleaseAge option to resolveCommand  -  by @​antfubot in #​85 (3106c)

    View changes on GitHub

v0.14.5

Compare Source

Patch Changes

• 🐞🩹 fall back right away when navigator.serviceWorker.register() fulfills without a registration (e.g. Playwright's serviceWorkers: 'block'), instead of throwing Cannot read properties of undefined (reading 'active') (by @​MFA-G in #​764)

• 🐞🩹 read scroll offsets from the main thread instead of caching them, so scripts such as GA4 observe scrolling after their initial read. (by @​Niek in #​766)

• 🐞🩹 define HTMLCollection in the web worker so x instanceof HTMLCollection no longer throws (e.g. gtag's user-provided-data DOM scan) (by @​gioboa in #​762)

v10.1.5

Compare Source

What's Changed

• fix: use registered root for relative downloads by @​lprnmns in #​610

New Contributors

• @​lprnmns made their first contribution in #​610

Full Changelog: fastify/fastify-static@v10.1.4...v10.1.5

v4.5.0

Compare Source

   🚀 Features

• Emit an explicit color level from codeToANSI  -  by @​EvanProgramming in #​1320 (b968e)
• transformers:
  • Add classActiveCode option to transformerNotationWordHighlight  -  by @​curiouscoder-cmd in #​1201 (1e968)
  • Add transformerRenderLineNumber  -  by @​Divyapahuja31 and DIVYA PAHUJA in #​1238 (02f05)

   🐞 Bug Fixes

• cli:
  • Fall back to 'text' when file/URL has no extension and no -…  -  by @​VeerShah41 in #​1248 (c00e5)
  • Fall back to plaintext for unknown languages  -  by @​Jinjiang and @​antfubot in #​1319 (4eb7c)
• primitive:
  • Dispose wasm memory for lazily embedded langs  -  by @​rgodha24 in #​1322 (b5423)
  • Dedupe grammar injections on repeated language loads  -  by @​antlio and @​antfubot in #​1318 (33678)
• stream:
  • Compile Solid renderer with babel-preset-solid  -  by @​BetterAndBetterII and @​antfu in #​1309 (8724e)
• transformers:
  • Keep notations on content comment lines  -  by @​tianrking in #​1308 (d25ab)
  • Handle split # and -- comments in rose-pine theme  -  by @​anandghegde in #​1317 (63e6d)
• twoslash:
  • Avoid crash when @​errors nodes land past EOF  -  by @​Yahiro025, @​cursoragent and @​antfubot in #​1316 (5c919)

   🏎 Performance

• engine-js: Reuse each regex's last search while tokenizing a line  -  by @​issacgerges and @​antfu in #​1314 (2ef5e)
• stream: Group tokens by line in the React renderer  -  by @​CNSeniorious000 in #​1311 (81200)
• vitepress-twoslash: Replace lz-string with lz-string-es  -  by @​antfubot in #​1324 (2e170)

    View changes on GitHub

v4.4.3

Compare Source

   🚀 Features

• Update deps and grammar  -  by @​antfu (72f26)

   🐞 Bug Fixes

• vitepress-twoslash: Delay twoslash tooltip show by 200ms  -  by @​antfubot in #​1306 (9f2aa)

    View changes on GitHub

v4.4.2

Compare Source

   🐞 Bug Fixes

• Unbreak CI after vite 8 bump and grammar/deps updates  -  by @​antfubot in #​1300 (c5d31)

   🏎 Performance

• core: Token split performance improvements  -  by @​amadeus in #​1301 (41896)

    View changes on GitHub

v4.4.1

Compare Source

   🚀 Features

• Update deps  -  by @​antfu (e07fd)

    View changes on GitHub

v4.4.0

Compare Source

   🚀 Features

• Update grammars and themes  -  by @​antfu (dec2a)

   🐞 Bug Fixes

• core: Guess yaml from frontmatter segments  -  by @​antfubot in #​1299 (1b8b6)

    View changes on GitHub

v4.3.1

Compare Source

   🚀 Features

• Add type field for ThemedToken  -  by @​ije in #​1293 (585b5)

    View changes on GitHub

v4.3.0

Compare Source

   🚀 Features

• stream: Add svelte renderer  -  by @​CNSeniorious000 and Anthony Fu (via agent) in #​1291 (2d18a)

    View changes on GitHub

v4.2.0

Compare Source

   🚀 Features

• Add @​shikijs/stream and @​shikijs/magic-move packages  -  by @​antfu in #​1283 (d031f)

   🐞 Bug Fixes

• transformers: Handle YAML comment prefixes correctly for v3  -  by @​AkaHarshit in #​1266 (f694a)
• vitepress-twoslash: Scroll blocking on mobile viewports  -  by @​micaiguai in #​1262 (9e0e8)

    View changes on GitHub

v4.1.0

Compare Source

   🐞 Bug Fixes

• twoslash: Forward tsModule to createTwoslasher  -  by @​arthurfiorette in #​1271 (be89a)

    View changes on GitHub

v0.14.5

Compare Source

Patch Changes

• fix: detect showLineNumbers wherever it appears in the meta string, not only before the highlight annotations, while ignoring the word when it is part of a character highlight such as /foo showLineNumbers bar/ (#​204)
• fix: stop character-highlight ranges leaking across patterns, so an id-only annotation like /foo/#a no longer makes a later pattern inherit the previous pattern's range, and a range-ignored occurrence no longer consumes the whole token it lives in (#​169)

Range-excluded occurrences are counted without changing the token tree, so unhighlighted markup remains intact.

Partial highlights contained within a transformer-generated element preserve that surrounding semantic element instead of cloning it.

v0.14.4

Compare Source

Patch Changes

• fix: prevent an infinite loop in inline character highlighting when a pattern matches text that spans multiple syntax tokens and occurs more than once on the same line (#​185)

v3.0.3

Compare Source

• Fix quote direction at the start of headings and table cells by @​geraschenko in #​103

v0.10.5

Compare Source

Patch changes

• Updated dependencies: satteri-mdxjs (Cargo)@​0.3.13

v0.10.4

Compare Source

Patch changes

• Updated dependencies: satteri-mdxjs (Cargo)@​0.3.13

v0.8.0

Compare Source

• [#​528] #​528
  Adding security policy
• [#​525] #​525
  Use Object.create for sourceContents
• [#​520] #​520
  Fixes types for SourceMapConsumer.initialize