llm: updates, fixes, refactor, ... #8684

haraldschilly · 2025-12-18T14:22:02Z

update all langchain code
refactor user defined models to common calls
- remove package deps we no longer need
add gemini 3, grok, update claude, etc.
expand tests
frontend: improve tokenizer, larger contexts

haraldschilly · 2025-12-19T16:37:24Z

@codex review

haraldschilly · 2025-12-19T17:12:42Z

@codex review

chatgpt-codex-connector

💡 Codex Review

cocalc/src/packages/frontend/account/useLanguageModelSetting.tsx

Lines 28 to 45 in cf43a76

    
           const haveOpenAI = useTypedRedux("customize", "openai_enabled"); 
        
           const haveGoogle = useTypedRedux("customize", "google_vertexai_enabled"); 
        
           const haveOllama = useTypedRedux("customize", "ollama_enabled"); 
        
           const haveCustomOpenAI = useTypedRedux("customize", "custom_openai_enabled"); 
        
           const haveMistral = useTypedRedux("customize", "mistral_enabled"); 
        
           const haveAnthropic = useTypedRedux("customize", "anthropic_enabled"); 
        
           const enabledLLMs: LLMServicesAvailable = useMemo(() => { 
        
             const projectsStore = redux.getStore("projects"); 
        
             return projectsStore.whichLLMareEnabled(project_id); 
        
           }, [ 
        
             haveOpenAI, 
        
             haveGoogle, 
        
             haveOllama, 
        
             haveCustomOpenAI, 
        
             haveMistral, 
        
             haveAnthropic, 
        
           ]);

xAI enablement never reaches account-level LLM picker

xAI was added as a new provider, but useLanguageModelSetting still only subscribes to the legacy *_enabled flags. Because xai_enabled is not read anywhere in this hook, enabledLLMs is memoized without the xAI flag and will never refresh when administrators turn xAI on. In environments where xAI is enabled (especially if it is the only enabled provider), components that rely on this hook (AI Formula, Jupyter AI helpers, chat, etc.) will continue to treat xAI as unavailable and fall back to some other default model.

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".

… timing wrapping

haraldschilly · 2025-12-22T17:00:11Z

@codex review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

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".

src/packages/server/llm/evaluate-lc.ts

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

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".

src/packages/server/llm/evaluate-lc.ts

…test coverage

haraldschilly · 2025-12-23T15:34:40Z

@codex review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

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".

src/packages/server/llm/evaluate-lc.ts

haraldschilly · 2025-12-23T15:49:35Z

@codex review

chatgpt-codex-connector

💡 Codex Review

Here are some automated review suggestions for this pull request.

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".

src/packages/server/llm/evaluate-lc.ts

haraldschilly added PR-work in progress PR-TODO-cocalc2 merge/migrate this PR into CoCalc2 in the future labels Dec 18, 2025

haraldschilly force-pushed the llm-upd-20251218 branch 18 times, most recently from e988b93 to 39dde7f Compare December 18, 2025 18:54

llm: updates, fixes, refactor, ...

7de83d6

haraldschilly force-pushed the llm-upd-20251218 branch 3 times, most recently from 45edc28 to cacdfe6 Compare December 19, 2025 13:30

server/llm: refactor user-models <-> langchain and add more testing

c3f5d73

haraldschilly force-pushed the llm-upd-20251218 branch from cacdfe6 to c3f5d73 Compare December 19, 2025 14:27

frontend/llm: improve client tokenizer for on-the-fly estimations

8c55be0

haraldschilly force-pushed the llm-upd-20251218 branch 2 times, most recently from d4a071e to 24034f4 Compare December 19, 2025 16:31

haraldschilly force-pushed the llm-upd-20251218 branch from ffd3032 to cf43a76 Compare December 19, 2025 17:12

chatgpt-codex-connector bot reviewed Dec 19, 2025

View reviewed changes

haraldschilly force-pushed the llm-upd-20251218 branch from cf43a76 to 49f0aaf Compare December 22, 2025 13:53

llm: bugfixing

c6484f8

haraldschilly force-pushed the llm-upd-20251218 branch 2 times, most recently from b0ef108 to b59ab38 Compare December 22, 2025 14:59

sagemathinc deleted a comment from chatgpt-codex-connector bot Dec 22, 2025

haraldschilly force-pushed the llm-upd-20251218 branch 3 times, most recently from 5f5d25a to 14afd81 Compare December 22, 2025 16:18

frontend/llm: switch token counting to estimations with tokenx; debug…

53eec99

… timing wrapping

haraldschilly force-pushed the llm-upd-20251218 branch from 14afd81 to 53eec99 Compare December 22, 2025 16:31

chatgpt-codex-connector bot reviewed Dec 22, 2025

View reviewed changes

src/packages/server/llm/evaluate-lc.ts Show resolved Hide resolved

chatgpt-codex-connector bot reviewed Dec 22, 2025

View reviewed changes

src/packages/server/llm/evaluate-lc.ts Show resolved Hide resolved

llm: update packages again (new release just happend); better ollama …

1534e6e

…test coverage

haraldschilly force-pushed the llm-upd-20251218 branch 5 times, most recently from d1ddc8d to 08b7d47 Compare December 23, 2025 15:16

chatgpt-codex-connector bot reviewed Dec 23, 2025

View reviewed changes

src/packages/server/llm/evaluate-lc.ts Show resolved Hide resolved

haraldschilly force-pushed the llm-upd-20251218 branch from 08b7d47 to 38a5245 Compare December 23, 2025 15:45

llm: validate user-defined model data and better typing/bugfix

81cf811

haraldschilly force-pushed the llm-upd-20251218 branch from 38a5245 to 81cf811 Compare December 23, 2025 15:48

haraldschilly marked this pull request as ready for review December 23, 2025 15:49

chatgpt-codex-connector bot reviewed Dec 23, 2025

View reviewed changes

src/packages/server/llm/evaluate-lc.ts Show resolved Hide resolved

	const haveOpenAI = useTypedRedux("customize", "openai_enabled");
	const haveGoogle = useTypedRedux("customize", "google_vertexai_enabled");
	const haveOllama = useTypedRedux("customize", "ollama_enabled");
	const haveCustomOpenAI = useTypedRedux("customize", "custom_openai_enabled");
	const haveMistral = useTypedRedux("customize", "mistral_enabled");
	const haveAnthropic = useTypedRedux("customize", "anthropic_enabled");

	const enabledLLMs: LLMServicesAvailable = useMemo(() => {
	const projectsStore = redux.getStore("projects");
	return projectsStore.whichLLMareEnabled(project_id);
	}, [
	haveOpenAI,
	haveGoogle,
	haveOllama,
	haveCustomOpenAI,
	haveMistral,
	haveAnthropic,
	]);

llm: updates, fixes, refactor, ... #8684

Are you sure you want to change the base?

llm: updates, fixes, refactor, ... #8684

Uh oh!

Conversation

haraldschilly commented Dec 18, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

haraldschilly commented Dec 19, 2025

Uh oh!

haraldschilly commented Dec 19, 2025

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

haraldschilly commented Dec 22, 2025

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

haraldschilly commented Dec 23, 2025

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

haraldschilly commented Dec 23, 2025

Uh oh!

chatgpt-codex-connector bot left a comment

Choose a reason for hiding this comment

💡 Codex Review

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

haraldschilly commented Dec 18, 2025 •

edited

Loading