benchflow 0.7.6

What's Changed

• Bump BenchFlow to 0.7.6.dev0 by @bingran-you in #1042
• fix(acp): modernize codex-acp dispatch — shim 1.6.0, effort via model id by @bingran-you in #1044
• Preserve healthy native ACP subscription results by @kywch in #1049
• Add --publish-bucket and --eval-results-* flags to bench eval run by @NathanHB in #1035
• feat(viewer): reviewer-grade trajectory pages, multi-run browsing, and hf:// dataset sources by @ljr145733 in #1034
• fix(deps): update RestrictedPython to 8.3 by @kywch in #1070
• Cap OpenClaw Claude output tokens by @kywch in #1075
• fix(cli): reject --trials > 1 without --matrix by @Benjamin-eecs in #1064
• Bump claude-agent-acp pin 0.40.0 -> 0.73.0 so claude-fable-5-1 can run by @yuchenlwu in #1086
• fix(sandbox): use the verifier-setup budget for hardening execs by @Benjamin-eecs in #1062
• fix(eval): re-run infra-retryable verifier-errored tasks on resume by @Benjamin-eecs in #1063
• fix(acp): cap how long a pending tool call defers the idle watchdog by @Benjamin-eecs in #1066
• Fix Gemini ACP model IDs with Google provider prefixes by @kywch in #1030
• fix(judge): parse native edit targets by @kywch in #1069
• Honor Codex proxy-owned model selection by @kywch in #1076
• fix(acp): preserve usage and terminal evidence on timeout by @kywch in #1080
• Add Z.ai Coding Plan routing by @kywch in #1074
• fix(litellm): provision Vertex sandbox runtime by @kywch in #985
• fix: guard _classify_completed_outcomes against non-dict rewards by @VaisakhiMishra in #1054
• Release BenchFlow 0.7.6 by @bingran-you in #1103

New Contributors

• @NathanHB made their first contribution in #1035
• @ljr145733 made their first contribution in #1034
• @Benjamin-eecs made their first contribution in #1064
• @yuchenlwu made their first contribution in #1086
• @VaisakhiMishra made their first contribution in #1054

Full Changelog: v0.7.5...v0.7.6