OpenAIReasoning modelGPT-5-Codex
Best configuration ranks #127 of 610 on the BenchLeader Index at 58.1 ±3.9. Last measured 15 Sept 2025stale: no new result in six months. Released 15 Sept 2025.
- Blended price
- $3.44/M
- $1.25 in · $10.00 out
- Output speed
- –
- First answer
- –
- Context
- 400k
- Overall index58
- Reasoning64
- Coding53
- Agents & tools56
- Knowledge61
- Instruction following72
- Multimodal58
- Long context62
- Composite60
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| high | – | – | – | – | $0.0035 | Agents & tools 45 · Coding 54 |
| defaultbest | 58.1 | #127 | – | – | $0.0035 | Agents & tools 56 · Coding 53 · Composite 60 · Instruction following 72 · Knowledge 61 · Long context 62 · Multimodal 58 · Reasoning 64 |
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | high | default | Source |
|---|---|---|---|
| GPQA Diamond (AA)not in index | – | 83.7%#153 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | – | 27.9%#123 | Artificial Analysis |
| Kagi LLM Benchmark | – | 70.3%#26 | Kagi LLM Benchmark |
Coding
Agents & tools
| Benchmark | high | default | Source |
|---|---|---|---|
| Terminal-Bench | – | 44.3%#33 | Terminal-Bench |
| APEX-Agents | 20.1%#37 | – | Mercor |
| Terminal-Bench Hard | – | 37.9%#61 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | – | 86.8%#82 | Artificial Analysis |
Knowledge
| Benchmark | high | default | Source |
|---|---|---|---|
| AA-Omniscience | – | -7.5#125 | Artificial Analysis |
Instruction following
| Benchmark | high | default | Source |
|---|---|---|---|
| IFBench | – | 74.2%#36 | Artificial Analysis |
Multimodal
| Benchmark | high | default | Source |
|---|---|---|---|
| MMMU-Pro | – | 73.8%#92 | Artificial Analysis |
Long context
| Benchmark | high | default | Source |
|---|---|---|---|
| AA-LCR | – | 71.7%#144 | Artificial Analysis |
Composite
| Benchmark | high | default | Source |
|---|---|---|---|
| AA Intelligence Index | – | 24.9#117 | Artificial Analysis |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.130 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0035 | $0.0032 | – |
| Summarise a 30-page report | 12,000 / 600 | $0.021 | $0.011 | – |
| Code edit | 6,000 / 1,500 | $0.022 | $0.018 | – |
| Agentic coding session | 60,000 / 4,000 | $0.115 | $0.065 | – |
| Structured extraction | 2,000 / 200 | $0.0045 | $0.0028 | – |
See also
Data as of 9 Sept 2026. Compare these configurations.