Revision History
AgentBench · 8 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: Corrected AgentBench's test-set size (1,014, not 1,360) and model counts by paper version (25/27/29), the claim that GLM-4 and Claude 3 were added in late 2023 (they arrived in the 2025 v3), fabricated OS/KG average figures, text-davinci-002's score and qualitative open-source scores (replaced with reported values), WebArena's date, the false claim that the FC release had no updated leaderboard, and unsupported 'direct successors' and survey attributions (plus 33 minor fixes)
Version 8 · Nov 5, 2026, 10:40 AM
- Improve internal links (2026-10-13)
Version 7 · Oct 13, 2026, 10:15 AM
- Added 6 contextual internal links
Version 6 · Jul 23, 2026, 03:27 PM
- Pre-update snapshot before enhancement to 4552 words.
Version 2 · May 18, 2026, 05:31 AM