Commits

Branch selector

 main

User selector

All users

All time

Commit History

Commits on Jan 30, 2026

• Add model normalization to /v1/responses endpoint to avoid LiteLLM mismatch warning

  AlexsanderHamir

  committed
  ab53af6

  Copy full SHA for ab53af6

• Normalize deployment model IDs (e.g. gpt-4o-mini-data-zone) to avoid LiteLLM mismatch warning

  AlexsanderHamir

  committed
  8a627f9

  Copy full SHA for 8a627f9

• Echo requested model in responses to avoid LiteLLM mismatch warning

  AlexsanderHamir

  committed
  758ba88

  Copy full SHA for 758ba88

• Add total_tokens to Anthropic-style usage responses for LiteLLM compatibility

  AlexsanderHamir

  committed
  e4b8dda

  Copy full SHA for e4b8dda

Commits on Jan 26, 2026

• Fix Bedrock response formats for Mistral and Cohere models

  Show description for 35c132c
  AlexsanderHamir

  committed
  35c132c

  Copy full SHA for 35c132c

• Add Gemini API auth support and missing Bedrock/Gemini endpoints

  Show description for 6169ebb
  AlexsanderHamir

  committed
  6169ebb

  Copy full SHA for 6169ebb

• Add Gemini API auth support and missing Bedrock/Gemini endpoints

  Show description for ac124a7
  AlexsanderHamir

  committed
  ac124a7

  Copy full SHA for ac124a7

Commits on Jan 17, 2026

• Fix KeyError: 'content' for Claude models in predict/rawPredict endpoints

  Show description for 5d39758
  AlexsanderHamir

  committed
  5d39758

  Copy full SHA for 5d39758

• Fix Gemini model detection: check request body for model info, remove debug prints, improve error messages to client

  AlexsanderHamir

  committed
  55ea80e

  Copy full SHA for 55ea80e

• Fix Gemini detection: also check extracted model name, not just path pattern

  AlexsanderHamir

  committed
  b90ccfd

  Copy full SHA for b90ccfd

• Simplify format detection: use path structure - Gemini paths get Gemini format, all others get Anthropic format with content field

  AlexsanderHamir

  committed
  9edb77a

  Copy full SHA for 9edb77a

• Add safety check for content field and improve model detection from request body

  AlexsanderHamir

  committed
  63dc256

  Copy full SHA for 63dc256

• Fix: detect Gemini models and return Gemini format, only default to Anthropic for simple paths when model unknown

  AlexsanderHamir

  committed
  73ccdce

  Copy full SHA for 73ccdce

• Force Anthropic format for all simple generateContent paths (not just when model undetected)

  AlexsanderHamir

  committed
  b37e9a7

  Copy full SHA for b37e9a7

• Fix Anthropic detection: ensure simple paths always default to Anthropic format

  AlexsanderHamir

  committed
  288687c

  Copy full SHA for 288687c

• Improve Anthropic model detection: check request body and default to Anthropic format for simple paths

  AlexsanderHamir

  committed
  c3def7f

  Copy full SHA for c3def7f

• Fix Vertex AI generateContent to return Anthropic Messages API format for Claude models

  AlexsanderHamir

  committed
  e620f13

  Copy full SHA for e620f13

• Add support for rawPredict Vertex AI endpoint

  AlexsanderHamir

  committed
  ce67cc2

  Copy full SHA for ce67cc2

• Add detailed error information to all error responses including request context, sanitized headers, and body preview

  AlexsanderHamir

  committed
  7ea2387

  Copy full SHA for 7ea2387

• Improve Vertex AI catch-all route: add comprehensive logging and pattern matching for debugging

  AlexsanderHamir

  committed
  636ec5c

  Copy full SHA for 636ec5c

Commits on Jan 16, 2026

• Fix Azure Responses API: handle list/dict input types and improve Vertex AI catch-all route

  AlexsanderHamir

  committed
  10b9c3e

  Copy full SHA for 10b9c3e

• Fix Vertex AI routes: remove broken colon routes, add catch-all handler, and improve error handling

  AlexsanderHamir

  committed
  14fa408

  Copy full SHA for 14fa408

• Add routes for :generateContent and :predict format to handle api_base:method calls

  AlexsanderHamir

  committed
  3cc0ff0

  Copy full SHA for 3cc0ff0

• Add Azure-compatible responses API endpoints (/openai/responses and /openai/v1/responses)

  AlexsanderHamir

  committed
  ac44197

  Copy full SHA for ac44197

• Add catch-all routes for Vertex AI endpoints to handle dynamic project/location/model paths

  AlexsanderHamir

  committed
  db04aa7

  Copy full SHA for db04aa7

Commits on Dec 2, 2025

• Add support for degraded providers.

  Show description for 90e87a4
  AlexsanderHamir

  committed
  90e87a4

  Copy full SHA for 90e87a4

Commits on Nov 19, 2025

• fix: make audio transcription lighter

  AlexsanderHamir

  committed
  d4f2bb3

  Copy full SHA for d4f2bb3

• fix: remove overhead from hot path

  AlexsanderHamir

  committed
  90ef9cf

  Copy full SHA for 90ef9cf

Commits on Nov 18, 2025

• fix: remove heavy overhead from fake endpoint

  AlexsanderHamir

  committed
  9d631f7

  Copy full SHA for 9d631f7

Commits on Nov 17, 2025

• fix: remove complexity from mock response

  Show description for fcbe159
  AlexsanderHamir

  committed
  fcbe159

  Copy full SHA for fcbe159

• fix: initialization order

  AlexsanderHamir

  committed
  77c570d

  Copy full SHA for 77c570d

• add: variation in output size return for audio/speech

  AlexsanderHamir

  committed
  5fa2340

  Copy full SHA for 5fa2340

Commits on Nov 16, 2025

• fix: audio endpoint to generate non-zero data & create larger payloads

  AlexsanderHamir

  committed
  5c691b6

  Copy full SHA for 5c691b6

Commits on Nov 15, 2025

• add support for /audio/transcriptions

  AlexsanderHamir

  committed
  f254f1b

  Copy full SHA for f254f1b

• adding support for /audio/speech

  AlexsanderHamir

  committed
  7c534a0

  Copy full SHA for 7c534a0

Pagination

Previous

Next