Skip to content

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
0.168%96.8%
The band covers the middle 80% of scores. The line marks the median.

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
151Gemini 3.5 FlashOpenRouter harness, google/gemini-3.5-flash-20260519, auto-routing, public 90-day aggregate· OpenRouter benchmark exports92.3%
152Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, Sail Research, public 90-day aggregate· OpenRouter benchmark exports92.3%
153Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, Modal, public 90-day aggregate· OpenRouter benchmark exports92.3%
154GPT-5.6 LunaOpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.92.3%
155Hy4 previewHF Eval Result· Reported by Tencent Hy· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.92.3%
156Qwen3.8 Maxxhigh· Epoch AI Benchmarking Hub data92.3%
157DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports92.3%
158Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, ParetoAlpha, public 90-day aggregate· OpenRouter benchmark exports92.3%
159Gemini 3.5 FlashOpenRouter harness, google/gemini-3.5-flash-20260519, Google AI Studio, public 90-day aggregate· OpenRouter benchmark exports92.3%
160Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, auto-routing, public 90-day aggregate· OpenRouter benchmark exports92.2%
161Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, Phala, public 90-day aggregate· OpenRouter benchmark exports92.2%
162Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, BaseTen, public 90-day aggregate· OpenRouter benchmark exports92.2%
163Kimi K3.1 ProOpenRouter harness, moonshotai/kimi-k3.1-pro-20261023, Wafer, public 90-day aggregate· OpenRouter benchmark exports92.2%
164GPT-5.6 SolOpenRouter harness, openai/gpt-5.6-sol-20260709, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports92.2%
165Kimi K3.1 ProOpenRouter harness, moonshotai/kimi-k3.1-pro-20261023, Makora, public 90-day aggregate· OpenRouter benchmark exports92.2%
166Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, KT, public 90-day aggregate· OpenRouter benchmark exports92.1%
167DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, DigitalOcean, public 90-day aggregate· OpenRouter benchmark exports92.1%
168GPT-6 SolOpenRouter harness, openai/gpt-6-sol-20260922, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports92.1%
169GPT-6 SolOpenRouter harness, openai/gpt-6-sol-20260922, OpenAI, public 90-day aggregate· OpenRouter benchmark exports92.1%
170Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, Moonshot AI, public 90-day aggregate· OpenRouter benchmark exports92.1%
171Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports92.1%
172Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, Chutes, public 90-day aggregate· OpenRouter benchmark exports92%
173Claude Opus 4.8OpenAI GPT-5.6 launch table· Reported by OpenAI· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.92%
174Claude Opus 4.8Qwen3.8 官方模型卡对照表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.92%
175Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, Alibaba, public 90-day aggregate· OpenRouter benchmark exports92%
176GPT-6 SolOpenRouter harness, openai/gpt-6-sol-20260922, auto-routing, public 90-day aggregate· OpenRouter benchmark exports91.9%
177Grok 4.7OpenRouter harness, x-ai/grok-4.7-20260916, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports91.9%
178Kimi K3.1 ProOpenRouter harness, moonshotai/kimi-k3.1-pro-20261023, Relace, public 90-day aggregate· OpenRouter benchmark exports91.9%
179Kimi K3high· Epoch AI Benchmarking Hub data91.9%
180DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports91.9%
181Gemini 3 Pro PreviewGemini 3 Pro Model Card · results as of November 2025 · no tools· Reported by Google DeepMind· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.91.9%
182Gemini 3 Pro PreviewOpus 4.6 System Card capability summary · adaptive thinking · max effort · default sampling · 5 trials unless noted· Reported by Anthropic· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.91.9%
183DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, NextBit, public 90-day aggregate· OpenRouter benchmark exports91.9%
184GPT-6.1 TerraOpenRouter harness, openai/gpt-6.1-terra-20261026, Amazon Bedrock, public 90-day aggregate· OpenRouter benchmark exports91.9%
185GLM 5.2max· Epoch AI Benchmarking Hub data91.9%
186Claude Sonnet 5.5OpenRouter harness, anthropic/claude-sonnet-5.5-20260928, Azure, public 90-day aggregate· OpenRouter benchmark exports91.9%
187Qwen3.8 MaxOpenRouter harness, qwen/qwen3.8-2.4t-a95b-20260812, Venice, public 90-day aggregate· OpenRouter benchmark exports91.8%
188DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, StreamLake, public 90-day aggregate· OpenRouter benchmark exports91.8%
189Nemotron 3.5 Super VLOpenRouter harness, nvidia/nemotron-3.5-super-vl-20261015, Io Net, public 90-day aggregate· OpenRouter benchmark exports91.7%
190Qwen 3.8 Flash NextQwen3.8-Flash-Next release/model card · published setting· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.91.7%
191Qwen3.8 FlashQwen3.8 Omni Flash 官方发布表· Reported by Qwen· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.91.7%
192DeepSeek V4 ProOpenRouter harness, deepseek/deepseek-v4-pro-20260423, Cloudflare, public 90-day aggregate· OpenRouter benchmark exports91.7%
193DeepSeek V4 Pro 0813max· Epoch AI Benchmarking Hub data91.7%
194MiniMax M3OpenRouter harness, minimax/minimax-m3-20260531, DeepInfra, public 90-day aggregate· OpenRouter benchmark exports91.6%
195MiniMax M3OpenRouter harness, minimax/minimax-m3-20260531, Minimax, public 90-day aggregate· OpenRouter benchmark exports91.6%
196GPT-5.6 Lunamax· Epoch AI Benchmarking Hub data91.6%
197DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, SiliconFlow, public 90-day aggregate· OpenRouter benchmark exports91.6%
198DeepSeek V4 Pro 0813OpenRouter harness, deepseek/deepseek-v4-pro-20260813, Baidu, public 90-day aggregate· OpenRouter benchmark exports91.6%
199Kimi K3OpenRouter harness, moonshotai/kimi-k3-20260715, AkashML, public 90-day aggregate· OpenRouter benchmark exports91.6%
200DeepSeek V4.1 Flash 1017OpenRouter harness, deepseek/deepseek-v4.1-flash-20261017, DekaLLM, public 90-day aggregate· OpenRouter benchmark exports91.6%

Showing 151–200 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)