github Agenta-AI/agenta v0.125.2

3 hours ago

What's Changed

  • feat(sdk): add Claude Haiku 5.5 to the model catalogs by @mmabrouk in #7821
  • [docs] Update API reference docs by @github-actions[bot] in #7824
  • fix(api): keep the partial answer after Stop and end the turn as stopped by @mmabrouk in #7827
  • fix(api): make the embed token scan linear by @mmabrouk in #7831
  • chore(deps): bump certifi from 2026.6.17 to 2026.10.18 in /services by @dependabot[bot] in #7825
  • [fix] Apply the runner output budget per step, not per turn by @mmabrouk in #7829
  • fix(services): install the mcp package litellm needs for MCP tools by @mmabrouk in #7830
  • fix(runner): keep a turn's work when a time limit ends it by @mmabrouk in #7823
  • chore(api): remove the llm-gateway-rollout and wallets-rollout PostHog flags by @mmabrouk in #7820
  • fix(frontend): keep a sent message when the user switches sessions by @mmabrouk in #7834
  • perf(api,sdk): run the agent turn-start lookups in parallel by @mmabrouk in #7832
  • fix(runner): make Haiku 5.5 run on Pi and list it in the picker by @mmabrouk in #7837
  • fix(frontend): keep a first message when the user opens another chat before it is sent by @mmabrouk in #7838
  • fix(api,frontend): let people write to channel chats in /m and keep private agents in channels by @mmabrouk in #7822
  • [release] v0.125.2 by @github-actions[bot] in #7826

Full Changelog: v0.125.1...v0.125.2

Don't miss a new agenta release

NewReleases is sending notifications on new releases.