Skip to main content
BenchLM
Advertise

Benchmark profile

GPQA Diamond (GPQA-D)

A display-only GPQA Diamond reference from provider comparison charts.

Data verified

Benchmark score on GPQA-D — July 23, 2026

BenchLM mirrors the published score view for GPQA-D. Sakana Fugu-Ultra leads the public snapshot at 95.5% , followed by Sakana Fugu (95.5%) and GPT-5.6 Sol (94.6%). BenchLM does not use these results to rank models overall.

1Closed

Sakana Fugu-Ultra

Sakana AI

sakana-fugu-ultra

95.5%
Overall —Context 1M
2Closed

Sakana Fugu

Sakana AI

sakana-fugu

95.5%
Overall —Context 1M
3Closed

GPT-5.6 Sol

OpenAI

gpt-5-6-sol

94.6%
Overall 81.96Context 1M
45 modelsKnowledgeCurrentDisplay onlyUpdated July 23, 2026

Benchmark score table (45 models)

Score
1
Sakana Fugu-UltraSakana AI · Closed
95.5%
2
Sakana FuguSakana AI · Closed
95.5%
3
GPT-5.6 SolOpenAI · Closed
94.6%
4
Gemini 3.1 ProGoogle · Closed
94.3%
5
Claude Opus 4.7 (Adaptive)Anthropic · Closed
94.2%
6
Claude Opus 4.8Anthropic · Closed
93.6%
7
GPT-5.5OpenAI · Closed
93.6%
8
Kimi K3Moonshot AI · Closed
93.5%
9
GPT-5.6 TerraOpenAI · Closed
92.9%
10
GPT-5.4OpenAI · Closed
92.8%
11
Gemini 3.5 FlashGoogle · Closed
92.7%
12
Qwen3.7 MaxAlibaba · Closed
92.4%
13
GPT-5.6 LunaOpenAI · Closed
92.3%
14
GLM-5.2Z.AI · Open weight
91.2%
15
Kimi K2.6Moonshot AI · Open weight
90.5%
16
Qwen3.7 PlusAlibaba · Closed
90.3%
17
DeepSeek V4 Pro (Max)DeepSeek · Open weight
90.1%
18
Interfaze BetaInterfaze · Closed
89.9%
19
Muse SparkMeta · Closed
89.5%
20
Claude Opus 4.6Anthropic · Closed
89.2%
21
DeepSeek V4 Pro (High)DeepSeek · Open weight
89.1%
22
Grok 4.20xAI · Closed
88.5%
23
DeepSeek V4 Flash (Max)DeepSeek · Open weight
88.1%
24
InklingThinking Machines Lab · Open weight
87.9%
25
Kimi K2.5Moonshot AI · Open weight
87.6%
26
DeepSeek V4 Flash (High)DeepSeek · Open weight
87.4%
27
Hy3 PreviewTencent · Open weight
87.2%
28
Nemotron 3 UltraNVIDIA · Open weight
87.0%
29
MiniMax M2.7MiniMax · Open weight
87.0%
30
GLM-5.1Z.AI · Open weight
86.2%
31
GLM-5Z.AI · Open weight
86.0%
32
MAI-Thinking-1Microsoft · Closed
84.2%
33
Gemma 4 12BGoogle · Open weight
78.8%
34
Trinity-Large-ThinkingArcee AI · Open weight
76.3%
35
DeepSeek V4 ProDeepSeek · Open weight
72.9%
36
Nemotron 3 Nano Omni 30B A3BNVIDIA · Open weight
72.2%
37
DeepSeek V4 FlashDeepSeek · Open weight
71.2%
38
ZAYA1-8BZyphra · Open weight
71.0%
39
Trinity-Large-PreviewArcee AI · Open weight
63.3%
40
Mellum2-12B-A2.5B-ThinkingJetBrains · Open weight
57.6%
41
ZAYA1-74B-PreviewZyphra · Open weight
57.3%
42
Soofi S 30B-A3BSoofi Project · Open weight
43.4%
43
Mellum2-12B-A2.5B-InstructJetBrains · Open weight
40.9%
44
MiniCPM5-1BOpenBMB · Open weight
26.3%
45
LFM2.5-230MLiquidAI · Open weight
25.4%

The published GPQA-D snapshot places Sakana Fugu-Ultra first at 95.5%. The third row is 0.9 points behind. The broader top-10 range is 2.7 points, so many of the published results sit in a relatively narrow band.

45 models have been evaluated on GPQA-D. The benchmark falls in the Knowledge category. This category carries a 12% weight in BenchLM.ai's overall scoring system. GPQA-D is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About GPQA-D

Year

2026

Tasks

Graduate-level science questions

Format

Multiple choice questions

Difficulty

Graduate level

BenchLM stores GPQA-D separately from the standardized GPQA row when providers publish exact chart values that should not overwrite the core weighted benchmark.

Trinity-Large-Thinking: Scaling an Open Source Frontier Agent

BenchLM freshness & provenance

Version

GPQA-D 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does GPQA-D measure?

A display-only GPQA Diamond reference from provider comparison charts.

Which model scores highest on GPQA-D?

Sakana Fugu-Ultra by Sakana AI currently leads with a score of 95.5% on GPQA-D.

How many models are evaluated on GPQA-D?

45 AI models have been evaluated on GPQA-D on BenchLM.

Compare Top Models on GPQA-D

Sakana Fugu-Ultra vs Sakana FuguSakana Fugu vs GPT-5.6 SolGPT-5.6 Sol vs Gemini 3.1 ProGemini 3.1 Pro vs Claude Opus 4.7 (Adaptive)
Last updated: July 23, 2026 · BenchLM version GPQA-D 2026

Choose a model with this week’s evidence

Join 2,000+ readers for ranking moves, pricing changes, and the claims that still need proof.

One email each week. Unsubscribe anytime.