Skip to content

feat(provider): adapt deepseek-v4-flash wire params and family detection - #1367

Closed
TheGreatAxios wants to merge 3 commits into
mainfrom
cl-10241-openai-compatible-adapter-deepseek-v4-flash-params
Closed

TheGreatAxios wants to merge 3 commits into
mainfrom
cl-10241-openai-compatible-adapter-deepseek-v4-flash-params

Conversation

@TheGreatAxios

Copy link
Copy Markdown
Collaborator

Summary

  • Adapts DeepSeek V4 Flash wire parameters in the OpenAI-compatible adapter, mapping the model native params to the adapter contract.
  • Extends provider family detection to recognize DeepSeek V4 Flash variants in provider-family.ts.

Verification

  • 15/15 adapter tests pass.
  • bun run typecheck, bun run build, and bun run test pass on this branch.

Fixes CL-10241

DeepSeek V4 Flash is the exception to the historical DeepSeek 400 rule:
its encoder keeps every turn's reasoning when tools are present, so
stripping it rendered prior tool turns empty and the model stopped
thinking. Add the deepseek-v4-flash family (provider-family.ts) and
apply the model-card wire shape for that family in the openai-compatible
adapter: chat_template_kwargs.thinking, effort folded onto low/high/max,
agentic temperature 1.0 / top_p 0.95, and stream_options.include_usage.
The family regex keeps other DeepSeek models on the old strip behavior.
Added 6 unit tests (15 total pass). CL-10241.
@linear-code

linear-code Bot commented Oct 9, 2026

Copy link
Copy Markdown

CL-10241

@TheGreatAxios
TheGreatAxios marked this pull request as ready for review October 9, 2026 03:02
@TheGreatAxios

Copy link
Copy Markdown
Collaborator Author

Superseded by the deepseek-v4-adapter rework: behavior merged via #1364/#1365/#1369 and the new per-family adapter (deepseek-v4-adapter.ts + registry) relocates the V4 quirks out of the generic openai-compatible adapter. See docs/plans/deepseek-v4-adapter-rework.md.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant