Benchmark
GPQA Diamond
Science benchmarks measure how well models answer science questions. Scores are percentages. Higher is better.
- Models and systems measured
- 282
- Scores recorded
- 1623
- Sources
- 4
- IQ weight
- 1.72
Score distribution
- 79.3%
- Mean
- 84.5%
- Median
- 15.7%
- Std dev
- 56.8%
- 10th percentile
- 92.2%
- 90th percentile
- 0.168%
- Min
This benchmark has a weight of 1.72 for chat models, measured over 272 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model or system | |
|---|---|---|
| 1351 | GPT OSS 20bofficial model card Table 3 · reasoning medium · with tools; Table 3 does not name the exact tool · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 67.1% |
| 1352 | Gemma 4 31B ITOpenRouter harness, google/gemma-4-31b-it-20260402, Venice, public 90-day aggregate· OpenRouter benchmark exports | 67% |
| 1353 | Llama 4 Maverick· Epoch AI Benchmarking Hub data | 67% |
| 1354 | GPT OSS 20bOpenRouter harness, openai/gpt-oss-20b, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports | 67% |
| 1355 | GPT 4.1· Epoch AI Benchmarking Hub data | 66.9% |
| 1356 | DeepSeek ChatOpenRouter harness, deepseek/deepseek-chat-v3, StreamLake, public 90-day aggregate· OpenRouter benchmark exports | 66.9% |
| 1357 | GLM 4.6vOpenRouter harness, z-ai/glm-4.6-20251208, auto-routing, public 90-day aggregate· OpenRouter benchmark exports | 66.8% |
| 1358 | qwen3-vl-30bOpenRouter harness, qwen/qwen3-vl-30b-a3b-instruct, Alibaba, public 90-day aggregate· OpenRouter benchmark exports | 66.8% |
| 1359 | Claude Sonnet 4· Epoch AI Benchmarking Hub data | 66.7% |
| 1360 | Gemini 2.5 Pro Preview 05-06· Epoch AI Benchmarking Hub data | 66.7% |
| 1361 | GPT 5.1none· Epoch AI Benchmarking Hub data | 66.7% |
| 1362 | GPT OSS 20bOpenRouter harness, openai/gpt-oss-20b, Phala, public 90-day aggregate· OpenRouter benchmark exports | 66.5% |
| 1363 | GLM 4.6vOpenRouter harness, z-ai/glm-4.6-20251208, Novita, public 90-day aggregate· OpenRouter benchmark exports | 66.3% |
| 1364 | Claude Sonnet 3.7· Epoch AI Benchmarking Hub data | 66% |
| 1365 | GPT OSS 20bofficial model card Table 3 · reasoning medium · no tools · pass@1· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 66% |
| 1366 | GLM 4.6vOpenRouter harness, z-ai/glm-4.6-20251208, Z.AI, public 90-day aggregate· OpenRouter benchmark exports | 66% |
| 1367 | Llama 4 MaverickOpenRouter harness, meta-llama/llama-4-maverick-17b-128e-instruct, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports | 66% |
| 1368 | Llama 4 MaverickOpenRouter harness, meta-llama/llama-4-maverick-17b-128e-instruct, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports | 66% |
| 1369 | Qwen3 235B A22B Instruct 2507OpenRouter harness, qwen/qwen3-235b-a22b-07-25, GMICloud, public 90-day aggregate· OpenRouter benchmark exports | 66% |
| 1370 | GPT 4.1OpenRouter harness, openai/gpt-4.1-2025-04-14, OpenAI, public 90-day aggregate· OpenRouter benchmark exports | 66% |
| 1371 | GPT-4.1 mini· Epoch AI Benchmarking Hub data | 65.8% |
| 1372 | Llama 4 MaverickOpenRouter harness, meta-llama/llama-4-maverick-17b-128e-instruct, auto-routing, public 90-day aggregate· OpenRouter benchmark exports | 65.8% |
| 1373 | qwen3-vl-30bOpenRouter harness, qwen/qwen3-vl-30b-a3b-instruct, PrimeIntellect, public 90-day aggregate· OpenRouter benchmark exports | 65.8% |
| 1374 | Qwen3-4B-Thinking-2507· OpenEvals Leaderboard data· Reported by the model's developer. The source relaying it did not verify it. | 65.8% |
| 1375 | Qwen3 32B· Epoch AI Benchmarking Hub data | 65.7% |
| 1376 | gemini-2.0-pro-exp-02-05· Epoch AI Benchmarking Hub data | 65.7% |
| 1377 | GPT OSS 20bOpenRouter harness, openai/gpt-oss-20b, auto-routing, public 90-day aggregate· OpenRouter benchmark exports | 65.6% |
| 1378 | QwQ Plus· Epoch AI Benchmarking Hub data | 65.4% |
| 1379 | QwQ 32B· Epoch AI Benchmarking Hub data | 65.3% |
| 1380 | GPT-4.1 miniOpenRouter harness, openai/gpt-4.1-mini-2025-04-14, auto-routing, public 90-day aggregate· OpenRouter benchmark exports | 65.2% |
| 1381 | Qwen3 30B A3B InstructOpenRouter harness, qwen/qwen3-30b-a3b-instruct-2507, auto-routing, public 90-day aggregate· OpenRouter benchmark exports | 65.2% |
| 1382 | Qwen3 30B A3B InstructOpenRouter harness, qwen/qwen3-30b-a3b-instruct-2507, Alibaba, public 90-day aggregate· OpenRouter benchmark exports | 65.2% |
| 1383 | DeepSeek R1OpenRouter harness, deepseek/deepseek-r1-0528, Novita, public 90-day aggregate· OpenRouter benchmark exports | 65.2% |
| 1384 | Nemotron 3.5 LightningOpenRouter harness, nvidia/nemotron-3.5-lightning-20260807, Darkbloom, public 90-day aggregate· OpenRouter benchmark exports | 65.1% |
| 1385 | Qwen3 32BOpenRouter harness, qwen/qwen3-32b-04-28, Groq, public 90-day aggregate· OpenRouter benchmark exports | 65% |
| 1386 | GPT-4.1 miniOpenRouter harness, openai/gpt-4.1-mini-2025-04-14, OpenAI, public 90-day aggregate· OpenRouter benchmark exports | 64.9% |
| 1387 | qwen3-vl-30bOpenRouter harness, qwen/qwen3-vl-30b-a3b-instruct, Novita, public 90-day aggregate· OpenRouter benchmark exports | 64.9% |
| 1388 | Llama 4 MaverickOpenRouter harness, meta-llama/llama-4-maverick-17b-128e-instruct, Novita, public 90-day aggregate· OpenRouter benchmark exports | 64.5% |
| 1389 | GPT OSS 20bOpenRouter harness, openai/gpt-oss-20b, Darkbloom, public 90-day aggregate· OpenRouter benchmark exports | 64.5% |
| 1390 | qwen3-vl-30bOpenRouter harness, qwen/qwen3-vl-30b-a3b-instruct, auto-routing, public 90-day aggregate· OpenRouter benchmark exports | 64.4% |
| 1391 | DeepSeek R1 Distill Qwen 32B· Epoch AI Benchmarking Hub data | 64.1% |
| 1392 | Gemini 2.0 Flash· Epoch AI Benchmarking Hub data | 64.1% |
| 1393 | GPT-5.4 mininone· Epoch AI Benchmarking Hub data | 64.1% |
| 1394 | Llama 4 MaverickOpenRouter harness, meta-llama/llama-4-maverick-17b-128e-instruct, Parasail, public 90-day aggregate· OpenRouter benchmark exports | 64.1% |
| 1395 | Qwen3 30B A3B InstructOpenRouter harness, qwen/qwen3-30b-a3b-instruct-2507, Ambient, public 90-day aggregate· OpenRouter benchmark exports | 64.1% |
| 1396 | GPT-4.1 miniOpenRouter harness, openai/gpt-4.1-mini-2025-04-14, Azure, public 90-day aggregate· OpenRouter benchmark exports | 64% |
| 1397 | Qwen3 14B· Epoch AI Benchmarking Hub data | 63.8% |
| 1398 | GPT 4.1OpenRouter harness, openai/gpt-4.1-2025-04-14, auto-routing, public 90-day aggregate· OpenRouter benchmark exports | 63.7% |
| 1399 | Qwen3 VL 235BOpenRouter harness, qwen/qwen3-vl-235b-a22b-instruct, Venice, public 90-day aggregate· OpenRouter benchmark exports | 63.7% |
| 1400 | GPT 4.1OpenRouter harness, openai/gpt-4.1-2025-04-14, Azure, public 90-day aggregate· OpenRouter benchmark exports | 63.7% |
Showing 1,351–1,400 of 1,621 results.
Not ranked
The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.
| Model or system | |
|---|---|
| MiMo V2 Flash Baseofficial Base table · 5-shot· Reported by Xiaomi MiMo· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 55.1% |
| Qwen3.8-2.4T-A95Blifecycle-page core-benchmark claim· Reported by Alibaba Cloud Model Studio· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it. | 92.6% |
Tokens the model generated per task, as measured by the source. Comparable within this benchmark only.
Data credits: Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links. · OpenRouter public benchmark leaderboard and search-lane exports, CC BY 4.0. Fractional source scores converted to percentages; immutable dated snapshots. (licence) · OpenEvals/leaderboard-data, MIT (licence)