Public KG record
Terminal-Bench 4.0
Recorded overview
General agent benchmark.
An evidence-backed entity from the project knowledge graph; detailed tool metadata is still being added.
Recorded details
- Benchmarks: used as professional coding evaluation showing Astra occupies cost-efficiency frontier majority
- Benchmarks: Claude Opus 5.5结果按xhigh努力等级报告
- Benchmarks: 用于评估 Astra，并比较启用与未启用水印的输出
- Benchmarks: M3.1 Flash得分38.1，Opus 5得分51.8
- Benchmarks: 测试编码 agent 解决命令行问题的能力
- Benchmarks: Mistral Large 4 得分 25%
- Performance: Astra 未启用水印为 53.90%，启用水印为 56.06%
- Use cases: Used to evaluate Mistral Large 4's agentic coding performance
Evidence coverage
The KG currently records 7 known information dimensions and has 10 undocumented dimensions, linked to 21 public source articles. These counts describe record coverage, not model performance or verified fit.
Related public records
- Astra — evaluated on
- Claude Fable 5.1 — evaluated on
- Claude Haiku 5.5 — evaluated on
- Claude Mythos 5.1 — evaluated on
- Claude Opus 5.5 — evaluated on