Skip to content

feat(ai-baseten): integrate Baseten adapter - #869

Open
khipu-agent wants to merge 1 commit into
profullstack:masterfrom
khipu-agent:feat/adapter-baseten
Open

feat(ai-baseten): integrate Baseten adapter#869
khipu-agent wants to merge 1 commit into
profullstack:masterfrom
khipu-agent:feat/adapter-baseten

Conversation

@khipu-agent

Copy link
Copy Markdown

Replaces the Baseten BYOK stub with a working OpenAI-compatible integration, following the same pattern as the Tier 1 adapters merged in #67-#71 (see #63).

  • generate() posts to Baseten Model APIs at https://inference.baseten.co/v1/chat/completions
  • Default model: deepseek-ai/DeepSeek-V4-Pro; models list: deepseek-ai/DeepSeek-V4-Pro, openai/gpt-oss-120b, zai-org/GLM-5.2, zai-org/GLM-5 (from Baseten's current Model APIs docs)
  • ctx.dryRun short-circuits before the network call
  • Errors include status code + first 200 chars of the response body, with the API key redacted
  • inputTokens/outputTokens populated from data.usage
  • Fixes the stub's malformed tokenSetup (secretKey was a URL instead of BASETEN_API_KEY)

Validation:

  • pnpm exec vitest run packages/ai/baseten/src/index.test.ts — 11 passed
  • pnpm --filter @profullstack/sh1pt-ai-baseten typecheck — clean
  • git diff --check — clean

Replace the BYOK stub with a real OpenAI-compatible chat completions
integration against Baseten Model APIs (https://inference.baseten.co),
matching the conventions of the integrated adapters: dry-run
short-circuit, usage token mapping, clean base URL validation, and
redacted error excerpts. Fixes the malformed tokenSetup secretKey.
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