Bump inspect-ai from 0.3.239 to 0.3.240

[dependabot]

Bumps inspect-ai from 0.3.239 to 0.3.240.

Changelog

Sourced from inspect-ai's changelog.

0.3.240 (15 June 2026)

• Anthropic: Support for server-side refusal fallback via the fallback_models generate config (Claude 5+ on the first-party Anthropic API).
• Anthropic: cache_ttl model arg for specifying the prompt cache TTL ("5m" or "1h").
• Anthropic: Support for web search dynamic filtering on Claude 4.6 and later models.
• Anthropic: Raise a clear error when reasoning_tokens is set on Claude 4.7+ or Claude 5 (which removed the budget_tokens control); use reasoning_effort instead.
• OpenAI: Forward GenerateConfig.extra_headers on the chat completions API path (previously only the responses and OpenAI-compatible paths honored it).
• OpenAI: Preserve Responses API NamespaceToolParam grouping through agent bridge.
• SageMaker: Fix streaming handler dropping reasoning tokens from thinking models.
• Model API: Record model fallbacks as a typed ModelOutput.fallback and roll them up per-sample on EvalSample.model_fallbacks (and sample summaries); expose a fallbacks count column in samples_df().
• Adaptive Connections: Raise the default minimum from 4 to 10.
• Control Channel: inspect eval / inspect eval-set now bind a per-process control server (AF_UNIX, default on) exposing a read surface for the live run. New inspect ctl commands let another process (CLI, scripts, agents) observe a running eval / eval-set.
• Transcript: Reuse a persistent per-thread SQLite connection in the realtime sample buffer database.
• Logging: Complete samples by logging the resident in-memory events directly rather than reading every event back out of the realtime buffer database.
• Logging: Re-enable realtime logging and score display for large runs (≥1000 samples) — the buffer-database and sample-completion improvements make them inexpensive enough to leave on.
• Tools: internal_tool_type() helper and INTERNAL_TOOL_TYPE options key marking server-side tools.
• Agent Bridge: Preserve source="operator" provenance on operator-injected messages.
• Task Display: Update log/plain progress after errors.
• Eval Logs: Support for writing to Hugging Face Storage Buckets.
• Sandbox tools: Inject tool support as a PyInstaller --onedir bundle instead of a single StaticX executable.
• Sandbox: self_check now verifies that non-ASCII (UTF-8) command output round-trips correctly on exec stdout/stderr.
• S3: Retry when requests have stale signatures.
• Inspect View: Display model fallbacks (samples grid column, has_fallbacks/fallbacks filter variables, sample header, transcript fallback marker and model-event badge).
• Inspect View: Fix occasional intermittent hang while attempting to navigate to sample
• Inspect View: Improve server side tool rendering transcripts and messages
• Inspect View: Fix stuck completed streaming samples
• Bugfix: Fix sample summary thinning mutating shared ChatMessage objects.
• Bugfix: Fix a rare OperationalError: unable to open database file when concurrent evals share a log directory.

Commits

• 9d08432 update changelog for release
• 64a2747 Bugfix: Fix a rare OperationalError: unable to open database file when conc...
• 29e1d63 docs: fix broken Eval Options link in options.qmd (#4252)
• 9c5d072 remove failing tests
• f24c527 docs: fix 404 on Evals page "Inspect Evals" doc links (#4251)
• 40f9d87 update changelog
• 6a961ec Preserve NamespaceToolParam through agent bridge (#4249)
• e8c49ef Adaptive Connections: Raise the default minimum from 4 to 10. (#4248)
• 42356db Anthropic: cache_ttl model arg for specifying the prompt cache TTL ("5m" or...
• 25f096c Update to latest (#4244)
• Additional commits viewable in compare view

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.

---

Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

• @dependabot rebase will rebase this PR
• @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
• @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
• @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
• @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)