Repository navigation
🤖 fix: make Fast mode follow the chat for every provider - #6830
Conversation
|
Preview deployment for your docs. Learn more about Mintlify Previews.
💡 Tip: Enable Automations to automatically generate PRs for you. |
|
@codex review |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
🛡️ Codex Security Review · Automatically triggeredSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: e9216f2e44
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
@codex review Fast-mode shortcut now ignores key repeats, like Ultrafast. Please take another look. |
|
Codex Review: Didn't find any major issues. Chef's kiss. Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
🛡️ Codex Security Review · Automatically triggeredSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
|
@codex review Added: a speed choice on the new-workspace page is now saved as the project/global default, like the thinking level, and existing chats fall back to it. Please take another look. |
🛡️ Codex Security Review · Automatically triggeredSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: aa0577713b
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
@codex review Addressed the explicit xAI tier precedence and continuous-compaction speed findings; replied on the backend-defaults thread (matches thinking level by design). Please take another look. |
🛡️ Codex Security Review · Automatically triggeredSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
|
Codex Review: Didn't find any major issues. Another round soon, please! Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
|
@codex review Added: Pro/Cyber reasoning mode on the new-workspace page is now saved as the project/global default too (same rules as speed and thinking level). Please take another look. |
🛡️ Codex Security Review · Automatically triggeredSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
|
Codex Review: Didn't find any major issues. Hooray! Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
Anthropic and xAI Fast mode still wrote provider-wide config, so the toggle followed the model. Use the chat's serviceTier pick for all Fast-mode providers, like the reasoning level, and map it to Anthropic speed / xAI service_tier.
…t the chat's speed
|
@codex review Rebased onto main (no conflicts); no code changes since the last review. |
6cde90f to
5c60f86
Compare
🛡️ Codex Security Review · Automatically triggeredSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 5c60f863ef
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: 5c60f863ef
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
|
Codex Review: Didn't find any major issues. Keep them coming! Reviewed commit: ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
If Codex has suggestions, it will comment; otherwise it will react with 👍. Codex can also answer questions or update the PR. Try commenting "@codex address that feedback". |
🛡️ Codex Security Review · Automatically triggeredSecurity review completed. No security issues were found in this pull request. Reviewed commit: Only the user who started this review can view the report in Codex. ℹ️ About Codex security reviews in GitHubThis is an experimental Codex feature. Security reviews are triggered when:
Once complete, Codex will leave suggestions, or a comment if no findings are found. |
This branch was successfully deployed
2 participants
Summary
Fast mode (and Ultrafast) now follows the same persistence and visibility rules as the reasoning level, for every provider (OpenAI, Anthropic, xAI):
Background
#6088 made OpenAI speed chat-scoped. Anthropic (
speed: "fast") and xAI (service_tier) were left out on purpose, so their toggle still wroteproviders.jsonc. As a result, the selector indicator, the shortcut, the palette and/fastread and wrote state that belonged to the provider, not to the chat. Switching a chat between Opus, Grok and GPT changed whether Fast showed as on, and toggling Fast on Opus in one chat turned it on in every Opus chat.Implementation
resolveChatServiceTier(provider, chatTier, providersConfig)is the one source of truth. The chat's choice ("priority"means Fast for any provider) comes first; a chat with no choice inherits the provider default (Anthropicspeed: "fast"or the provider'sserviceTier). The selector, the global shortcut and/fast//ultrafastall read it and write throughsetWorkspaceServiceTier, the same path reasoning uses. That path keeps an unsent choice until the next send and nudges the active turn. On a creation draft it saves to the draft record./sidechat (getSideChatPaneWorkspaceId, the lookuppaneHandlesKeyEventuses), else the routed chat or creation draft. Key repeats are ignored, so holding the chord produces one change.applyFastModeToggle, the restore-target writers,isFastModeActive) and their tests. Settings → Providers still controls the provider default.providerModelFactory). The chat's choice already reaches the factory (as the OpenAI tier, plus the livegetServiceTierOverride). The factory now reads it before non-OpenAI routes remove it:speed: "fast"is pinned from the chat's choice, or from the provider setting if there is none. A mid-turn change applies to the next call through the same live-override path OpenAI uses.priority/defaultand applied per request. An explicitproviderOptions.xai.serviceTierstill wins, including over live changes.readChatServiceTier,resolveAnthropicFastMode,resolveXAIServiceTier) socreateModelCoreEffect's complexity (CC) does not grow.AgentSession's send-options-to-provider-options helper moved tocommon/utils/ai/chatServiceTier.tsand is shared with the summarizer.Validation
speedin both directions.speedbetween calls.service_tierfollows the chat's choice on each request, and an explicit xAI tier wins.getSideChatPaneWorkspaceIdis covered next to the existingpaneHandlesKeyEventtests./fasttests:tests/uithinking-selector flow checks that Fast turned on in a Grok chat is still on after switching back to GPT. It could not run locally because Node 20 lacksPromise.withResolvers. CI runs it.Risks
Medium-low.
Generated with
xum• Model:anthropic:claude-opus-5-5• Thinking:high