Skip to content

v0.3.290

Choose a tag to compare

@github-actions github-actions released this 14 Nov 21:14
e9a8055

0.3.290 (2026-11-13)

Features

  • active_model_event_id() returns the in-flight model call attempt's event id (#6529) (c060c20)
  • each message keeps one id through the agent bridge and the log, and tool events record the message that proposed them (#6143) (187d19f)
  • excluded epochs are left out of epoch reduction, and score reasons gain excluded and zeroed (#6528) (b4aa05b)
  • google: support Gemini 4 Carbon (#6466) (8c6f22a)
  • google: support Gemini 4 Hydrogen (#6547) (d0c2ae1)
  • logs of errored or cancelled evals keep their total sample count (#6476) (5a1654f)
  • meta: support Muse Ember 1.0 and Muse Wonder 1.0 (#6467) (e95b39b)
  • mistral: support Mistral Large 4 (#6531) (1ac1482)
  • sandbox agent bridge inside a sandbox identity scope serves only the scope's user (#6270) (9a0b77e)
  • sentinel: count a sentinel sidecar's judging time as waiting and its judge models' usage in the eval (#6513) (0152514)
  • sentinel: sign calls to a sentinel sidecar instead of sending the eval's token (#6444) (9961043)

Bug Fixes

  • a bridged agent's host tools make no model call after a limit is reached (#6500) (55e9344)
  • a bridged agent's queued requests and native compactions make no model call after a limit is reached (#6434) (10dc86e)
  • a large denied sandbox bridge host tool call no longer uses extra memory to check for retries (#6471) (6a1efac)
  • a retry chosen in the cancel dialog no longer re-runs a task cancelled with inspect ctl (#6457) (40d60b8)
  • a second eval() in the same process ignores a changed display setting (#6267) (674ac5e)
  • ACP approval and ask_user requests stuck on a client that switched sample or rejected the request (#6487) (acd87bb)
  • agents wrapped with as_tool or as_solver fail when reused with limits (#6387) (842a233)
  • an operator cancel during scoring no longer hides a scorer error (#6455) (4364a56)
  • anthropic: long requests no longer retry forever on the 10 minute read timeout (#6511) (fbb88c6)
  • async fsspec reads block the event loop (#6012) (099d83a)
  • computer: type action doesn't use Return for newlines (#6490) (fba8be6)
  • count a sample once toward fail_on_error when its score_on_error scoring also fails (#6507) (36294bb)
  • eval() raises an error instead of silently running a different task registered under the same name (#6231) (918246a)
  • keep a host's Rich progress context usable after an eval (#6409) (95024b3)
  • keep the other batch results when one result row fails to parse (#6217) (36eaa3e)
  • list_files() no longer runs a path starting with - as a find action (#6489) (88eef6a)
  • math() timeouts no longer look like wrong answers in eval logs (#6384) (16a39eb)
  • messages_df() keeps messages that several samples share (#6389) (8abad63)
  • moonshot: honor low and high reasoning effort for Kimi K3 (#5946) (bc9998c)
  • openai-responses: preserve content_filter stop reason on partial tool calls (#6227) (d08cf15)
  • openai: computer tool results with no screenshot no longer fail the next request with a 400 (#6516) (35fa403)
  • pick the JSON reader from the URL path, not the whole URL (#6209) (3391d54)
  • read digits before a Chinese unit as one number (#6208) (c8c58d4)
  • recovered samples keep their store (#6165) (13149f6)
  • score numbers wrapped in full-width punctuation (#6381) (a82e702)
  • sentinel: a sidecar call that ends a sample on a token, cost or turn limit shows only the model's time (#6542) (e4435ba)
  • stop Hugging Face provider sending two BOS tokens for Gemma and Llama 3 (#6385) (2ce28a3)
  • vLLM and SGLang server API keys are no longer written to logs or startup errors (#6493) (793f79e)
  • web_search() Google provider no longer logs the API key on connection retry (#6491) (050c34f)

Performance Improvements

  • copy shard sample members into the merged log without recompressing them (#6488) (26f6d05)