GPQA Diamond leaderboard

Graduate-level questions in biology, physics and chemistry that experts answer well and search does not. Gemini 3.1 Pro Preview leads with 95.45%.

Models
120
Leader
Gemini 3.1 Pro Preview
Source
Vals AI
As of
Oct 24, 2026
  1. 1Gemini 3.1 Pro Preview95.5%
  2. 2GPT-5.6 Sol95.2%
  3. 3Grok 4.694.7%
  4. 4Gemini 3.8 Flash94.4%
  5. 5Gemini 3.7 Flash93.9%
  6. 6Qwen3.8-Max93.7%
  7. 7Gemini 3.6 Flash93.4%
  8. 8Claude Opus 593.4%
  9. 9Claude Fable 5.193.4%
  10. 10Claude Fable 593.2%
GPQA Diamond, accuracy against cost per test
Open weightsProprietaryBest for the price
20%40%60%80%100%$.1$1$10Cost per test, log scale →Gemini 3.1 Pro PreviewMiniMax-M3GPT-5.6 LunaGLM-5.3-Flashgpt-oss-20bGPT-5 nano

Full leaderboard

RankModelLabAccuracyCost / testHarnessPrice in / out
1Gemini 3.1 Pro PreviewGoogle95.45%$2 / $12—$2 / $12
2GPT-5.6 SolOpenAI95.20%$4 / $20—$4 / $20
3Grok 4.6xAI94.70%$2 / $6—$2 / $6
4Gemini 3.8 FlashGoogle94.44%$1.5 / $7.5—$0.82 / $4.13
5Gemini 3.7 FlashGoogle93.94%$1.5 / $7.5—$0.75 / $3.75
6Qwen3.8-MaxAlibaba93.69%$2 / $6—$2 / $6
7Gemini 3.6 FlashGoogle93.43%$1.5 / $7.5—$0.75 / $3.75
8Claude Opus 5Anthropic93.43%$5 / $25—$5 / $25
9Claude Fable 5.1Anthropic93.43%$10 / $50—$10 / $50
10Claude Fable 5Anthropic93.18%$10 / $50—$10 / $50
11GPT-5.5OpenAI93.18%$5 / $30—$5 / $30
12Kimi K3Moonshot AI92.93%$3 / $15—$0.24 / $11
13Grok 4.5xAI92.93%$2 / $6—$2 / $6
14Gemini 3.5 FlashGoogle92.68%$1.5 / $9—$1.50 / $9
15MiniMax-M3MiniMax92.68%$0.6 / $2.4—$0.30 / $1.20
16Claude Opus 4.8Anthropic92.42%$5 / $25—$5 / $25
18GPT-5.6 LunaOpenAI91.67%$0.2 / $1.2—$0.20 / $1.20
19Gemini 3 ProGoogle91.67%$2 / $12——
20GPT-5.2OpenAI91.67%$1.75 / $14—$1.75 / $14
21GPT-5.4OpenAI91.67%$2.5 / $15—$2.50 / $15
22Grok 4.3xAI91.41%$1.25 / $2.5—$1.25 / $2.50
23Muse Spark 1.1Meta91.16%$1.25 / $4.25—$1.25 / $4.25
24GPT-5.6 TerraOpenAI90.91%$2 / $12—$2 / $12
25Qwen3.7-Max-PreviewAlibaba90.15%$2.5 / $7.5—$1.48 / $4.42
26Claude Opus 4.7Anthropic90.15%$5 / $25—$5 / $25
27DeepSeek-V4-FlashDeepSeek89.90%$0.44 / $1.32—$0.042 / $0.18
28Muse SparkMeta89.65%N/A——
29Claude Opus 4.6Anthropic89.65%$5 / $25—$5 / $25
30DeepSeek-V4-ProDeepSeek89.39%$1.32 / $3.96—$0.66 / $1.98
31Kimi K2.6Moonshot AI89.14%$0.95 / $4—$0.65 / $3.41
32Claude Sonnet 5Anthropic88.89%$2 / $10—$2 / $10
33Qwen3.8-27BAlibaba88.89%$0.5 / $3—$0.20 / $2.50
34Grok 4.20xAI88.64%$2 / $6—$1.25 / $2.50
35Grok 4xAI88.13%$3 / $15——
36GLM-5.3Z.ai88.13%$1.4 / $4.4—$0.079 / $1.40
37Gemini 3 FlashGoogle87.88%$0.5 / $3—$0.50 / $3
38Qwen 3.5 PlusAlibaba87.37%$0.4 / $2.4—$0.30 / $1.80
39Qwen3.6-PlusAlibaba87.37%$0.5 / $3—$0.33 / $1.95
40InklingThinky87.12%$1 / $4.05—$1 / $4.05
41GPT-5.1OpenAI86.62%$1.25 / $10—$1.25 / $10
42MiniMax-M2.7MiniMax86.62%$0.3 / $1.2—$0.30 / $1.20
43GLM-5.3-FlashZ.ai86.36%$0.075 / $0.25—$0.15 / $0.50
44Nemotron 3 UltraNVIDIA86.11%N/A—$0.60 / $2.40
46GPT-5OpenAI85.61%$1.25 / $10—$1.25 / $10
47GLM-5.2Z.ai85.61%$1.4 / $4.4—$0.034 / $7
48Claude Sonnet 4.6Anthropic85.61%$3 / $15—$3 / $15
50Ling 3.0 FlashinclusionAI84.85%$0.075 / $0.22—$0.021 / $0.063
52GLM-5.1Z.ai84.52%$1 / $3.2—$1.40 / $4.40
53Grok 4.1 FastxAI84.34%$0.2 / $0.5——
54o3OpenAI84.09%$2 / $8—$2 / $8
55Kimi K2.5Moonshot AI84.09%$0.6 / $3—$0.50 / $2.50
56Gemini 3.5 Flash-LiteGoogle83.84%$0.3 / $2.5—$0.30 / $2.50
57Inkling-SmallThinky83.59%$0.3 / $1.2—$0.45 / $1.20
58GLM-5Z.ai83.33%$1 / $3.2—$1 / $3.20
59GPT-5.4 miniOpenAI83.08%$0.75 / $4.5—$0.75 / $4.50
60Qwen3.5-FlashAlibaba82.83%$0.1 / $0.4—$0.065 / $0.26
61MiMo-V2.5-ProXiaomi82.58%$0.435 / $0.87—$0.30 / $0.61
62MiniMax-M2.5MiniMax82.07%$0.3 / $1.2—$0.30 / $1.20
63Claude Sonnet 4.5Anthropic81.63%$3 / $15—$3 / $15
65MiMo-V2.5Xiaomi81.57%$0.14 / $0.28—$0.12 / $0.24
66Gemini 3.1 Flash-Lite PreviewGoogle81.06%$0.25 / $1.5—$0.25 / $1.50
67Gemini 2.5 ProGoogle80.81%$1.25 / $10—$1.25 / $10
68GPT-5 miniOpenAI80.30%$0.25 / $2—$0.25 / $2
70GLM-4.7Z.ai80.05%$0.6 / $2.2—$0.54 / $1.98
71Claude Opus 4.5Anthropic79.55%$5 / $25—$5 / $25
74gpt-oss-120bOpenAI78.54%$0.15 / $0.6—$0.15 / $0.60
75MiniMax-M2.1 PreviewMiniMax78.54%$0.3 / $1.2—$0.30 / $1.20
76Kimi K2 ThinkingMoonshot AI78.54%$0.6 / $2.5—$0.60 / $2.50
77Qwen3-Max-PreviewAlibaba77.78%$1.2 / $6——
78GPT-5.4 nanoOpenAI77.53%$0.2 / $1.25—$0.20 / $1.25
81DeepSeek-V3.2-ExpDeepSeek76.26%$0.56 / $1.68—$0.28 / $0.42
82o3-miniOpenAI75.50%$1.1 / $4.4—$1.10 / $4.40
85o4-miniOpenAI74.50%$1.1 / $4.4—$1.10 / $4.40
86GLM-4.6Z.ai74.50%$0.6 / $2.2—$0.55 / $2.20
87Grok 3 PreviewxAI74.24%$3 / $15——
88o1OpenAI73.23%$15 / $60—$15 / $60
89Grok 3 Mini ReasoningxAI72.98%$0.3 / $0.5——
90Claude Haiku 4.5Anthropic72.22%$1 / $5—$1 / $5
91GLM-4.5Z.ai72.22%$0.6 / $2.2—$0.60 / $2.20
92Claude Opus 4Anthropic71.72%$15 / $75——
93Kimi K2 InstructMoonshot AI71.46%$1 / $3——
95Qwen3-235B-A22BAlibaba70.20%$0.22 / $0.88—$0.09 / $0.55
96Claude Opus 4.1Anthropic69.95%$15 / $75—$15 / $75
97Llama 4 MaverickFireworks AI69.44%$0.22 / $0.88—$0.19 / $0.65
98Claude Sonnet 4Anthropic69.44%$3 / $15——
99gpt-oss-20bOpenAI68.94%$0.07 / $0.3—$0.018 / $0.09
101GPT-4.1 miniOpenAI67.93%$0.4 / $1.6—$0.40 / $1.60
102Claude 3.7 SonnetAnthropic67.42%$3 / $15——
103GPT-4.1OpenAI65.40%$2 / $8—$2 / $8
104Gemini 2.0 FlashGoogle65.15%$0.1 / $0.4——
105Grok 4.1 Fast Non-ReasoningxAI65.15%$0.2 / $0.5——
106Gemini 2.5 Flash PreviewGoogle64.14%$0.1 / $0.4——
107GPT-5 nanoOpenAI63.38%$0.05 / $0.4—$0.05 / $0.40
109Magistral MediumMistral AI62.37%$2 / $5——
110Grok 4 FastxAI62.12%$0.2 / $0.5——
112Claude 3.5 SonnetAnthropic59.34%$3 / $15——
113MiMo-V2-FlashXiaomi59.34%$0.1 / $0.3——
114Gemini 1.5 ProGoogle58.33%$1.25 / $5——
116Gemini 2.5 Flash Preview 4/17Google57.58%$0.3 / $2.5——
117Laguna XS.2Poolside55.05%N/A——
118DeepSeek-V3DeepSeek54.55%$0.9 / $0.9—$0.29 / $1.14
120GPT-4.1 nanoOpenAI50.76%$0.1 / $0.4—$0.10 / $0.40
121Mistral SmallMistral AI50.76%$0.2 / $0.6——
122Grok 2xAI50.76%$2 / $10——
123GPT-4oOpenAI50.25%$5 / $15—$2.50 / $10
124Llama 3.3 Instruct TurboMeta50.00%$0.88 / $0.88——
125Command ACohere48.48%$2.5 / $10—$0.30 / $1.50
126Mistral LargeMistral AI47.73%$2 / $6—$0.50 / $1.50
127Llama 3.3 Nemotron SuperMeta47.73%N/A——
128Llama 4 ScoutMeta46.97%$0.18 / $0.59—$0.10 / $0.30
129Gemini 1.5 FlashGoogle45.96%$0.075 / $0.3——
130Magistral Small 1.2Mistral AI44.19%$0.075 / $0.3——
131GPT-4o miniOpenAI44.19%$0.15 / $0.6—$0.15 / $0.60
132Claude 3.5 HaikuAnthropic37.88%$0.8 / $4——
133Jamba 1.6 LargeAI21 Labs36.11%$2 / $8——
134Mistral Medium 3.5Mistral AI34.85%$1.5 / $7.5—$1.50 / $7.50
135Jamba 1.6 MiniAI21 Labs32.07%$0.2 / $0.4——
136Command R+Cohere31.06%$2.5 / $10—$0.15 / $0.60
137GPT-3.5 TurboOpenAI30.56%$0.5 / $1.5—$1 / $2
138Laguna M.1Poolside27.02%N/A——

Related leaderboards

Sources

  • Vals AI, as of Oct 24, 2026
  • Provider list prices, as of Oct 24, 2026

Every number on this page comes from these sources; models without a published number show none. Download the data. Cite as “Nexus Explorer, nexus.inc/models”.