whatmodel.sh / GPT-5 nano
GPT-5 nano
↗ share
+ contribute
GPT-5 nano is a closed-weights language model from OpenAI, released August 2025.
$ how it compares
The LLMs ranked either side of GPT-5 nano, out of 828 with a score, then more from OpenAI.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 208 |  A.X K1 (519B-A33B) | 62±1 | local | ||
| 209 | gpt-oss-120b | 62±1 | api $0.26/M +local | ||
| 210 | GPT-5 nano | 62±1 | api $0.14/M | ||
| 211 |  Qwen3-VL 32B Thinking | 62±1 | api $0.28/M +local | ||
| 212 |  Gemini 2.0 Flash-Lite (001) | 62±2 | comparison only | ||
| more from OpenAI | |||||
| 1 | GPT-6 Astra | 87±1 | api $20.00/M | ||
| 3 | GPT-6.1 Sol | 85±1 | api $4.00/M | ||
| 7 | GPT-6 Sol | 83±1 | api $4.00/M | ||
$ what it's good at
agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #172 of 300 · 20% of its score · 6 tests
Beats o1 (2024-12-17), o3-mini, DeepSeek R1 0528 here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, Jamba 1.5 Large here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Terminal-Bench 2.0 with Terminus 2 agent (benchmark authors' runs) | 7.9% ✎ | ▃··▃·▃▃██▅▃▅ beats 1/14 | 4.3% |
| BrowserGym WorkArena-L1 (AgentLab GenericAgent) | 40.6% ✎ | █···█▄▄·▄█·█ beats 3/10 | 3.8% |
| BrowserGym WorkArena-L2 (AgentLab GenericAgent) | 3.4% ✎ | ▃▅▃·▅▃▃··█·▃ beats 3/11 | 3.6% |
| τ²-bench Telecom lab | 35.5% ✎ | ▃·▅▅▃▃··█▃▃▅ beats 2/13 | 3.2% |
| BFCL V4 (December 2025 archive) | 51.4% ✎ | ▁▁▁▂██▅▅▆▂▂▂ beats 59/77 | 3.1% |
| BrowserGym MiniWoB (AgentLab GenericAgent) | 64.8% ✎ | ▅····▅▅▃▃·▃█ beats 5/11 | 2.1% |
chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #173 of 393 * · 6.1% of its score · 1 test
Beats Seed-OSS 36B Instruct here, though they rank higher overall. Trails LFM2.5 1.2B Instruct, LFM2.5 1.2B Thinking, Gemma 3 12B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| LMArena (text) | 1337.5 ✎ | ▁▁▁▂▃▃▃▆▆▇█▃ beats 176/379 | 6.1% |
math▁▂▃▅▆▅▄█▄▁▁▁ #130 of 398 · 34% of its score · 7 tests
Beats Claude Sonnet 4.6, MiniMax-M3, MiMo-V2.5 (310B-A15B) here, though they rank higher overall. Trails Olmo 3.1 32B Think, Qwen3-VL 30B-A3B Thinking, Granite 4.2 8B here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| FrontierMath tiers 1–3 v2 | 20.0% ✎ | ▂··▁▁▃▄▄█▅▃▃ beats 14/82 | 12% |
| MATH level 5 | 94.9% ✎ | ▂▂▅▆▄█▆▃▅▂▃▄ beats 82/92 | 8.6% |
| OTIS mock AIME 2024–2025 | 81.1% ✎ | ▃▃▄▃▃▃▇▇█▃▂▅ beats 116/193 | 6.3% |
| MATH-Vision | 62.2% ✎ | ▁▁·▁··▄▁▃▇▇█ beats 7/24 | 2.2% |
| MathNet-Solve-Test micro-average accuracy (6,400 olympiad problems) | 42.2% ✎ | ▃▃···█▃·▃·▃▃ beats 4/8 | 2.0% |
| FrontierMath tier 4 v2 | 2.4% ✎ | ▁▃▁▂█▅▂▂▁▁▁▁ beats 7/63 | 1.6% |
| FormalProofBench (200 graduate-level Lean 4 problems, agentic) | 12.5% ✎ | ▃▅·▃··▅▃█··▃ beats 4/10 | 0.7% |
vision & documents▁▁▃▂▁▃▄▂▆▇█▄ #93 of 147 * · 6.4% of its score · 2 tests
Beats DeepSeek V4 Flash Vision (Exp) here, though they rank higher overall. Trails GPT-4o (2024-11-20), Mistral Small 4 (119B-A6.5B), Qwen3-VL 30B-A3B Thinking here, though they rank lower overall.
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Arena vision (style controlled) | 1146.2 ✎ | ▁▁▂▂▂▂▄▅▅▆██ beats 47/145 | 4.2% |
| MATH-Vision | 62.2% ✎ | ▁▁·▁··▄▁▃▇▇█ beats 7/24 | 2.2% |
other tests11 not in a task · 36% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| GPQA Diamond | 69.4% ✎ | ▁▁▁▁▂▄▆▄▅▇█▅ beats 104/208 | 11% |
| KMMLU-HARD (Daekeun Kim, zero-shot) | 51.7% ✎ | █···▁▁▁▂▃·▆▂ beats 13/23 | 6.5% |
| CLIcK Korean (Daekeun Kim) | 72.0% ✎ | ▆▄▄▄█▄▂·▄█·▄ beats 9/23 | 6.5% |
| HAE-RAE 1.0 (Daekeun Kim) | 78.6% ✎ | ▄▄▄▂▂█▂▄▂█▂▄ beats 10/22 | 5.2% |
| SimpleQA Verified | 11.7% ✎ | ▁▁▃▂▃▃▆▆█▃▂▅ beats 3/78 | 1.6% |
| SimpleQA lab | 11.0% ✎ | ▄▄▄▆█▆███·▂█ beats 13/32 | 1.3% |
| Chess Puzzles (Epoch) | 27.0% ✎ | █▁▁▁▁▂▁▄▂▂▁▁ beats 100/135 | 1.3% |
| Mystery Game Puzzles (Epoch) | 8.0% ✎ | █··▂▃▇▄█▂▃▁▁ beats 8/74 | 1.0% |
3 more tests
| test | result | vs the others in its run | counts |
|---|---|---|---|
| Open PT-LLM Leaderboard average (9 Portuguese tasks) | 79.8% ✎ | ▃▁▁▁▁▁▁▁▂█▅▁ beats 257/276 | 0.9% |
| Vectara hallucination leaderboard, 2025 dataset (HHEM-2.3, factual consistency rate) | 95.3% ✎ | ▁▁▂▃▄▄█▆▃▂▂▁ beats 69/140 | 0.2% |
| Vectara hallucination leaderboard (HHEM-2.3, factual consistency rate) | 89.5% ✎ | ▁▂▁▂▅█▂▃▄▃▁▁ beats 39/98 | 0.0% |
$ where to run it
Through an API
api · $0.14/M · $0.05 in / $0.4 out per 1M tokens · 400k context · platform.openai.com ↗ · price sourceStandard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ.
$ how the score was worked out · Epoch AI — internal evaluatio… 44%, Daekeun Kim Korean LLM benchm… 18%, LMArena leaderboard dataset 10% and 10 more
Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works
$ improve this page
Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.
data updated · sources and licences · json