Model
GPT-5 Codex
OpenAI · closed weights
Coding
CodingAgentic / Tool Use
Agentic / Tool Use| Benchmark | Score | Metric | Harness | Source | Confidence | Link |
|---|---|---|---|---|---|---|
| Terminal-Bench | 44.3 | accuracy | Codex CLI | Independent | medium | tbench.ai |
| Terminal-Bench | 43.4 | accuracy | Terminus 2 | Independent | medium | tbench.ai |
| Terminal-Bench | 41.3 | accuracy | Mini-SWE-Agent | Independent | medium | tbench.ai |
Reproduced by the community
Loading community runs…