Agent Diff Bench
Benchmark for evaluating agents on Slack, Linear, Box, Calendar via Bash & Python
- Domain
- rl-env
- License
- unknown
- Published
- Feb 2026
Cite
Notes
Only stored in your browser.
Top score 74.7% by DeepSeek V3.2 - 6 models reporting (2 frontier)
Score history
3Top models
6
Related tools
1Implementations, trainers, datasets and scaffolds linked to this eval.
FAQ
- What is Agent Diff Bench?
- Benchmark for evaluating agents on Slack, Linear, Box, Calendar via Bash & Python
- What is the current top score on Agent Diff Bench?
- The top reported score is 74.7% by DeepSeek V3.2, across 6 models reporting (2 from frontier labs).
- How can a model improve its Agent Diff Bench score?
- Tools linked to Agent Diff Bench on Sophon include DIFF Bench RL Env (Community) - RL environments, datasets, and scaffolds that target this eval.
- What license is Agent Diff Bench under?
- Agent Diff Bench is available under unknown.