OpenAIReasoning modelGPT-5 mini
Best configuration ranks #172 of 610 on the BenchLeader Index at 55.8 ±5.0. Last measured 17 Feb 2026stale: no new result in six months. Released 7 Aug 2025.
- Blended price
- $0.688/M
- $0.250 in · $2.00 out
- Output speed
- 103 tok/s
- First answer
- 74 s
- first token 6.81 s
- Context
- 400k
- Overall index56
- Reasoning56
- Coding48
- Agents & tools56
- Knowledge50
- Instruction following73
- Multimodal59
- Long context63
- Composite51
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 |
|---|---|---|---|---|---|---|
| minimal | 43.9 | #428 | 102 tok/s | 0.94 s | $0.0007 | Agents & tools 46 · Composite 42 · Instruction following 47 · Knowledge 39 · Long context 45 · Maths 36 · Multimodal 43 · Reasoning 40 |
| low | – | – | 103 tok/s | 74 s | $0.0007 | Maths 37 · Reasoning 34 |
| medium | 54.5 | #196 | 95 tok/s | 13 s | $0.0007 | Agents & tools 48 · Coding 49 · Composite 55 · Instruction following 69 · Knowledge 59 · Long context 58 · Maths 64 · Multimodal 53 · Reasoning 43 |
| high | 52.3 | #250 | 103 tok/s | 74 s | $0.0007 | Coding 55 · Human preference 57 · Knowledge 50 · Maths 52 · Multimodal 53 · Reasoning 48 |
| thinking | – | – | 103 tok/s | 74 s | $0.0007 | Instruction following 54 |
| defaultbest | 55.8 | #172 | 103 tok/s | 74 s | $0.0007 | Agents & tools 56 · Coding 48 · Composite 51 · Instruction following 73 · Knowledge 50 · Long context 63 · Multimodal 59 · Reasoning 56 |
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 | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| GPQA Diamond | 71.7%#150 | – | 71.7%#152 | 75.0%#138 | – | – | Epoch AI Benchmarking Hub |
| Humanity's Last Exam | – | – | – | – | – | 19.4%#18 | Scale AI / CAIS |
| LMArena Hard Prompts | – | – | – | 1403#150 | – | – | LMArena |
| GPQA Diamond (AA)not in index | 68.7%#297 | – | 80.3%#187 | – | – | 82.8%#163 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 5.0%#374 | – | 15.9%#186 | – | – | 21.5%#154 | Artificial Analysis |
| GPQA Diamond (Vals)not in index | – | – | – | 80.3%#65 | – | – | Vals AI |
| Kagi LLM Benchmark | – | – | – | – | – | 70.3%#26 | Kagi LLM Benchmark |
| ARC-AGI-1 | 5.3%#177 | 26.3%#149 | 37.3%#136 | 54.3%#118 | – | – | ARC Prize |
| ARC-AGI-2 | 1.7%#148 | 0.8%#164 | 4.0%#127 | 4.4%#125 | – | – | ARC Prize |
Coding
| Benchmark | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| SWE-bench Verified (Epoch) | – | – | 64.7%#27 | – | – | – | Epoch AI Benchmarking Hub |
| SciCode | – | – | – | – | – | 39.2%#114 | SciCode |
| WeirdML | – | – | – | 52.7%#60 | – | – | WeirdML |
| LMArena Coding | – | – | – | 1431#152 | – | – | LMArena |
| SciCode (AA)not in index | – | – | – | – | – | 39.0%#119 | Artificial Analysis |
| LiveCodeBench | – | – | – | 86.6%#20 | – | – | Vals AI |
| IOI | – | – | – | 6.8%#38 | – | – | Vals AI |
| SWE-bench (Vals)not in index | – | – | – | 60.8%#73 | – | – | Vals AI |
| SWE-bench Verified (bash only) | – | – | 59.8%#24 | – | – | 56.2%#27 | SWE-bench |
| SWE-bench Verified (any scaffold)not in index | – | – | 59.8%#32 | – | – | – | SWE-bench |
Agents & tools
| Benchmark | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | – | 31.9%#45 | – | – | 34.8%#42 | Terminal-Bench |
| Terminal-Bench Hard | 14.4%#189 | – | 28.8%#115 | – | – | 33.3%#90 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 31.9%#243 | – | 71.0%#148 | – | – | 68.4%#156 | Artificial Analysis |
| BFCL Overall | – | – | – | – | – | 55.5%#15 | Berkeley Function Calling Leaderboard |
Maths
| Benchmark | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | 6.0%#94 | 18.3%#80 | – | 46.7%#40 | – | – | Epoch AI Benchmarking Hub |
| FrontierMath Tier 4 | – | – | – | 12.2%#47 | – | – | Epoch AI Benchmarking Hub |
| OTIS Mock AIME | 55.6%#157 | – | 78.3%#106 | 86.7%#71 | – | – | Epoch AI Benchmarking Hub |
| MATH Level 5 | – | – | 96.8%#8 | 97.8%#3 | – | – | Epoch AI Benchmarking Hub |
| ProofBench | – | – | – | 9.0%#50 | – | – | Vals AI |
| AIME (Vals) | – | – | – | 91.5%#24 | – | – | Vals AI |
| MGSM | – | – | – | 92.6%#15 | – | – | Vals AI |
| MathArena Apex | – | – | – | 1.0%#35 | – | – | MathArena |
Knowledge
| Benchmark | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | 21.6%#61 | – | – | Epoch AI Benchmarking Hub |
| AA-Omniscience | -53.0#341 | – | -11.0#149 | – | – | -17.3#172 | Artificial Analysis |
| MMLU-Pro | – | – | – | 82.2%#75 | – | – | Vals AI |
| LegalBench | – | – | – | 81.8%#73 | – | – | Vals AI |
| CorpFin | – | – | – | 60.2%#70 | – | – | Vals AI |
| TaxEval | – | – | – | 75.2%#21 | – | – | Vals AI |
| MedQA | – | – | – | 96.1%#6 | – | – | Vals AI |
| MultiNRC | – | – | – | – | – | 23.9%#30 | Scale AI SEAL |
Instruction following
| Benchmark | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| IFBench | 45.6%#199 | – | 71.2%#60 | – | – | 75.4%#31 | Artificial Analysis |
| MultiChallenge | – | – | – | – | 59.0%#11 | – | Scale AI SEAL |
Human preference
| Benchmark | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | 1389#143 | – | – | LMArena |
Multimodal
| Benchmark | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| LMArena Vision | – | – | – | 1202#72 | – | – | LMArena |
| MMMU-Pro | 58.4%#184 | – | 68.8%#128 | – | – | 70.1%#116 | Artificial Analysis |
| VISTA | – | – | – | – | – | 50.4%#9 | Scale AI SEAL |
Long context
| Benchmark | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| Fiction.LiveBench 120k | – | – | 62.5%#13 | – | – | – | Fiction.live |
| AA-LCR | 39.3%#291 | – | 72.3%#137 | – | – | 72.3%#137 | Artificial Analysis |
Composite
| Benchmark | minimal | low | medium | high | thinking | default | Source |
|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | – | 145.5#69 | Epoch AI Benchmarking Hub |
| AA Intelligence Index | 9.9#315 | – | 20.6#169 | – | – | 17.4#199 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| OpenAI | 78 tok/s | 4.02 s | $0.250 | $2.00 | 400k | – |
| Azure | 51 tok/s | 6.81 s | $0.250 | $2.00 | 400k | – |
| OpenAI Flex | 33 tok/s | 15 s | $0.125 | $1.00 | 400k | – |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
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 $0.025 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0007 | $0.0006 | 1.3 min |
| Summarise a 30-page report | 12,000 / 600 | $0.0042 | $0.0022 | 1.3 min |
| Code edit | 6,000 / 1,500 | $0.0045 | $0.0035 | 1.5 min |
| Agentic coding session | 60,000 / 4,000 | $0.023 | $0.013 | 1.9 min |
| Structured extraction | 2,000 / 200 | $0.0009 | $0.0006 | 1.3 min |
See also
Data as of 9 Sept 2026. Compare these configurations.