Changelog
Features
- 621bdc8 feat(authkeys): import LiteLLM virtual keys by hash (#1130)
- 1e891f9 feat(cli): add gomodel migrate litellm config converter (#1109)
- 2083c6f feat(cli): import LiteLLM keys, teams, and budgets from its database (#1132)
- 439c868 feat(config): add secret references with env and file schemes (#1123)
- c45c370 feat(ext): admin-guarded extension routes and health checks (#1125)
- 04dafb8 feat(mcp): add virtual MCP servers (#1120)
- d26539f feat(openai): count Messages tokens exactly with OpenAI's input_tokens endpoint (#1117)
- f8c8b77 feat(storage): encrypt dashboard-managed secrets at rest (#1124)
Bug Fixes
- 2a64952 fix(chatgpt): forward prompt_cache_key and session-id to the Codex backend (#1145)
- 95a5067 fix(chatgpt): keep streamed output items in non-streaming responses (#1147)
- a63d4b1 fix(chatgpt): send prompt_cache_key as session-id when no session is detected (#1146)
- d22f9ec fix(models): retrieve models by any ID inference accepts and page the Anthropic list (#1118)
- a07fbbb fix(virtualmodels): skip disabled targets instead of failing the request (#1137)
Full Changelog: v0.1.99...v0.1.100