Oct 21 Today 09:20 Document intended and actual output-limit evidence in all v2/v3 READMEs Update 11 dataset READMEs and 11 trace-statistics READMEs with sampled historical context, response budgets, thinking and sampling settings. Explain the Terminus max_tokens forwarding bug and mark unverified fallback limits unknown. Add 11 evidence JSON files. Preserve task data, traces and reported scores. Oct 17 4 days ago 20:22 crosscodeeval_v2: refreshed release with current v1 tasks, uniform card and trace statistics Oct 16 5 days ago 09:09 Rename v2 selection assets folders to metadata Move eight selection manifests to metadata and update reproduction scripts and README references. 07:49 Simplify README wording and put rule-based checks last Use kept instead of retained in README prose. Order per-stage breakdowns as build/setup, no-op, reference solution, then rule-based checks. Preserve task counts, validation findings and schema field names. Oct 15 6 days ago 21:48 Use consistent short dataset names with v1 and v2 folders 20:55 Publish verified compact successful and failed v2 teacher traces 20:51 Publish TermiGen v2 and reuse v1 excluded task payloads 20:43 Publish crosscodeeval_v2 filtered by both models pass@16 results Retain 863 of 6710 v1 tasks solved by Qwen3-30B Instruct or Thinking; archive 5847 newly unsolved tasks alongside 1053 prior exclusions. Preserve all task payloads, reuse pinned v1 images, and expose per-model success counts and selection examples. 
