Skip to content

v1.1.16

Choose a tag to compare

View all tags
@github-actions github-actions released this 06 Jul 05:52
· 32 commits to main since this release
v1.1.16
This tag was signed with the committer’s verified signature.
MitulShah1 Mitul Shah
GPG key ID: 9A023A4CFFD1D25D
Verified
Learn about vigilant mode.
c9c4e12
This commit was created on GitHub.com and signed with GitHub’s verified signature.
GPG key ID: B5690EEEBB952194
Verified
Learn about vigilant mode.

[1.1.16] — 2026-07-06

Local & prediction-API provider fidelity release. The seventh provider-readiness remediation phase aligns the Hugging Face, Ollama, Ollama Cloud, and Replicate providers — which use task-specific, prediction, and native (non-OpenAI-wire) APIs — on request/response correctness. No breaking API changes relative to v1.1.15.

Fixed

  • Hugging Face default endpoint: the default base URL now targets https://router.huggingface.co/v1; the previous api-inference.huggingface.co host is retired and returns HTTP 410.
  • Hugging Face embeddings and image generation: both now use Hugging Face's task-specific routes and schemas (feature extraction returns a bare vector array; text-to-image returns raw image bytes) instead of OpenAI-shaped requests that hit the wrong endpoints.
  • Replicate token usage & finish reasons: chat responses now populate token usage from prediction metrics and normalize finish reasons instead of hardcoding them.
  • Ollama Cloud sampling parameters: stop/seed/presence/frequency penalties are now forwarded, remaining unsupported parameters are reported rather than dropped silently, finish reasons are normalized, and tool-call IDs are populated.
  • Ollama embedding dimensions: the dimensions parameter is now forwarded to Ollama's native embeddings endpoint.

Changed

  • Replicate authentication (operator-visible): the provider now sends the documented Bearer authorization scheme instead of the deprecated Token scheme (Replicate still accepts both).
  • Replicate request timeout: a tuned transport preset raises the response-header timeout so Prefer: wait predictions (held up to ~60s) are not aborted by the default 30s timeout.
  • Ollama Cloud embeddings: Ollama Cloud now supports an embeddings endpoint.
  • Base-URL validation is now enforced at construction for Hugging Face, Ollama, and Replicate.

Contributors

Thanks to everyone who shipped this release:

Full changelog: https://github.com/ferro-labs/ai-gateway/blob/v1.1.16/CHANGELOG.md

Contributors

MitulShah1
Assets 8
Loading