Kimi Release Notes

Polish ​

• web: Let wide Markdown tables grow beyond the reading column up to 1040px, scrolling horizontally inside the table when wider.
• web: Keep the server access token for up to 7 days across tab close and browser restarts, instead of asking for it again with every new tab.
• web: Add workspaces by typing an absolute path directly in the workspace picker's search box, with live validation and completion suggestions.
• web: Auto-enable the default thinking effort when switching to a model that supports effort levels in the web UI.
• Recognize the support_efforts and default_effort fields when importing a custom registry, so thinking effort levels are available for those models.
• Update the WebBridge install page link opened from the /plugins panel.
• Add a subagent.timeout_ms config option (or the KIMI_SUBAGENT_TIMEOUT_MS env var) to control how long a single subagent may run before timing out; the default is raised from 30 minutes to 2 hours.
• Add a print-mode background policy: set [background].print_background_mode = "steer" to keep kimi -p alive across background-task completions, so the main agent can be steered into follow-up turns.

Bug Fixes ​

• web: Fix sessions getting stuck in a sending state after a reconnect; turns that finish while the connection is down now stop the spinner and let the next message send normally. …

Polish ​

• Retry provider 429, overload, and other transient errors more reliably, honoring the server Retry-After delay, and surface retries in -p --output-format stream-json.

Bug Fixes ​

• Stop unsupported image formats (AVIF, BMP, TIFF, ICO, …) from breaking sessions at every entry point — including remote image URLs and images mislabeled by a tool — and recover an already-stuck session by dropping the offending image and retrying, so one such image can no longer make every later request fail.
• web: Fix the "Turn finished" desktop notification and completion sound firing twice per turn.
• web: Hide the internal image-compression note so it no longer renders as user message text.

Features ​

• web: Add notifications when a tool needs approval, and improve notification reliability.

Polish ​

• web: Polish the chat UI with Inter typography, localized labels, and tighter composer and menu styling.
• web: Polish the session sidebar layout, colors, icons, and typography.
• Display the Extra Usage (fuel pack) balance in the /usage and /status commands.
• Add a Kimi WebBridge entry to the Official tab of the /plugins panel that opens the WebBridge install page in your browser.

Bug Fixes ​

• Keep image-heavy sessions within provider request-size limits: oversized images (model-read and pasted, including WebP) are downscaled and compressed, HEIC/HEIF reads are refused with a platform-matched conversion command instead of poisoning the session, and an HTTP 413 request-too-large now recovers automatically — the request and /compact retry with older media replaced by text markers. The limits are configurable via [image] in config.toml (or KIMI_IMAGE_* env vars), and each core keeps its own settings so reloading one client's config no longer changes another client's compression.
• Fix resuming sessions whose original working directory no longer exists.
• Fix prompt-mode goals so they run until completion and report invalid goal commands before sending prompts. …

⚡ HighSpeed is now available ​

Kimi Code now offers two tiers — Standard and HighSpeed — built on the same model with identical coding ability; HighSpeed delivers roughly 5–6× the output speed of Standard. Switch instantly with /model in the CLI, or set the model ID to kimi-for-coding-highspeed in third-party tools. HighSpeed requires an Allegretto plan or above. → How to Switch Models

Extra Usage is now available ​

When your subscription quota runs out, keep making requests with your Extra Usage balance — it bypasses the monthly / weekly / hourly membership quota limits, works whenever any limit is reached, only deducts the Extra Usage balance, and doesn't affect your membership quota refresh. Only subscribed members can enable it, at Kimi homepage → Settings → Membership Plan → My Quota. → About Extra Usage

Kimi Code CLI

Features ​

• Add the Vercel plugin to the bundled plugin marketplace. Run /plugins and select Vercel Plugin to install it.

Bug Fixes ​

• Fix kimi -p runs exiting with code 0 when a turn fails.
• Prevent autonomous goals from being paused by model-reported status updates.
• Count the turn that starts an autonomous goal toward its turn budget.
• Raise the image downscale cap from 2000px to 3000px, and fix swapped width/height for EXIF-rotated (portrait) photos in compression captions and media read notes so region readback coordinates map correctly.
• web: Fix the connection error toast lingering after the WebSocket reconnects when returning from the background.
• Fix console windows flashing on Windows each time a hook runs.

Polish ​

• web: Redesign the scheduled reminder UI.
• web: Show session skills in the slash menu as /skill:<name> so they are distinguishable from built-in commands; typing the bare skill name still works.
• web: The composer model switcher switches the active session's model as before and additionally bumps the global default model, so new sessions inherit the choice.
• web: Press Enter to confirm in archive and other confirmation dialogs.
• Tighten goal-mode guidance for blocked and complete status updates. …

Bug Fixes ​

• Fix kimi -p abandoning background subagents that start late or run long, so their results reach the main agent.
• web: Recover chat streaming after a stale background-tab WebSocket instead of requiring a page refresh.
• Fix some third-party models (e.g. Opus 4.8) falling back to the family default max output tokens; an unrecognized minor now reuses the nearest earlier known version's limit.
• Honor explicit Anthropic max_output_size settings instead of clamping them to built-in ceilings.
• Stop showing tool-produced <system> metadata in tool outputs; failed tools now show their own error text.
• Fix goal completion and blocked updates to produce one final user-facing outcome summary from the tool result.
• Fix goal startup and queue handling so failed starts restore permission mode and queued goals wait behind new user messages.
• Fix goal token budgets to count model completion tokens and stop without extra continuation steps when the budget is exhausted.
• Fix goal tools being unavailable to the main agent, and return clear messages for invalid goal-control calls.
• Respect the --skills-dir flag in interactive mode.
• web: Fix several slash commands and skills not working on the new-session screen: /goal <objective> and slash skill activations (for example /pre-changelog) silently did nothing, and /btw [<question>] opened an empty side chat.

Polish ​ …

Features ​

• web: Add an Archived sessions page in Settings to browse and restore archived sessions. Open Settings → Archived to find it.
• Add experimental on-demand tool loading (select_tools) under the tool-select flag: a supporting model loads MCP tools only when needed instead of sending all of them in every request, preserving the provider prompt cache. Off by default and only active on models that declare the select_tools capability.

Bug Fixes ​

• Fix sessions that exist on disk but were missing from the session list or returned 404 on direct access, by rebuilding the session index at server startup.
• Fix the Bash and Edit tool cards collapsing, jumping, or flickering in height when results stream in or finish with short output, and visually separate the Bash command from its output.
• Fix the input box shifting upward after the slash command menu closes.
• Fix the edit approval preview shown by Ctrl+E to include surrounding context lines, matching the summary panel.
• Fix @ file completion missing deeply nested files in large projects after adding extra workspace directories.
• web: Fix several web layout and animation glitches: the collapsed sidebar now hides correctly, the chat history no longer replays its entrance animation when opening a session, and tool components no longer jump the conversation when expanded or collapsed. …