Reasoning · Benchmark profile

SuperGPQA

Graduate-level Google-proof Q&A expansion beyond GPQA Diamond for harder knowledge-reasoning items.

Ranking weightedCurrent / rollingActive8 models

Data verified 15 Jul 2026 · Methodology 2.3.0

Benchmark score on SuperGPQA

Leaderboard · 8 models

Claude Opus 4.6 leads at 95%, followed by Claude Sonnet 4.6 (95%) and Qwen3.7-Max (73.6%).

Ranking · 2.5%ReasoningCurrent

#1Closed

Claude Opus 4.6

Anthropic

claude-opus-4-6

95%

Overall #30 · Context 1M tokens

#2Closed

Claude Sonnet 4.6

Anthropic

claude-sonnet-4-6

95%

Overall #49 · Context 1M tokens

#3Closed

Qwen3.7-Max

Alibaba Cloud

qwen3.7-max

73.6%

Overall #18 · Context 1M tokens

ViewModelsColours

AnthropicAlibaba CloudMoonshot AIZ.AI

Lumina

SuperGPQA leaderboard

Accessible data for SuperGPQA leaderboard
Model	Rank	Model ID	Provider	Score	Unit
Claude Opus 4.6	#1	claude-opus-4-6	Anthropic	95	percent
Claude Sonnet 4.6	#2	claude-sonnet-4-6	Anthropic	95	percent
Qwen3.7-Max	#3	qwen-3-7-max	Alibaba Cloud	73.6	percent
Qwen3.6 Plus	#4	qwen3-6-plus	Alibaba Cloud	71.6	percent
Qwen3.7-Plus	#5	qwen-3-7-plus	Alibaba Cloud	71.4	percent
Claude Opus 4.5	#6	claude-opus-4-5	Anthropic	70.6	percent
Kimi K2.5	#7	kimi-k2-5	Moonshot AI	69.2	percent
GLM-5	#8	glm-5	Z.AI	66.8	percent

Benchmark score table (8 models)

One best compatible score per canonical product · higher is better

SuperGPQA leaderboard ranked by score
Rank	Model	Provider	License	Evidence use	Score
#1	Claude Opus 4.6 claude-opus-4-6	Anthropic	closed	Source carrier	95%
#2	Claude Sonnet 4.6 claude-sonnet-4-6	Anthropic	closed	Source carrier	95%
#3	Qwen3.7-Max qwen3.7-max	Alibaba Cloud	closed	Source carrier	73.6%
#4	Qwen3.6 Plus qwen3-6-plus	Alibaba Cloud	closed	Source carrier	71.6%
#5	Qwen3.7-Plus qwen3.7-plus	Alibaba Cloud	closed	Source carrier	71.4%
#6	Claude Opus 4.5 claude-opus-4-5	Anthropic	closed	Source carrier	70.6%
#7	Kimi K2.5 kimi-k2.5	Moonshot AI	open	Source carrier	69.2%
#8	GLM-5 glm-5	Z.AI	open	Source carrier	66.8%

The top of this snapshot is led by Claude Opus 4.6 at 95%; third place is 21.4 points behind. The top-8 spread is 28.2 points.

Configuration-preserving evidencePublished source rows (8)Exact source labels, configurations, versions and dated observations. Same-snapshot canonical/configuration projections are shown once.

Every distinct dated source-published row for SuperGPQA, including exact model configuration labels
Exact published model / configuration	Canonical product	Score	Version / system	Evidence	Source / observed
Exact model variant as listed on the BenchLM public benchmark leaderboard page. source label without registered configuration ID	Claude Opus 4.6	95%	CurrentBenchLM aggregated public evaluation	Source carrier · no additional scoring weight source-checked	BenchLM public benchmark leaderboards ↗Observed 2026-07-15 · checked 2026-07-15
Exact model variant as listed on the BenchLM public benchmark leaderboard page. source label without registered configuration ID	Claude Sonnet 4.6	95%	CurrentBenchLM aggregated public evaluation	Source carrier · no additional scoring weight source-checked	BenchLM public benchmark leaderboards ↗Observed 2026-07-15 · checked 2026-07-15
Exact model variant as listed on the BenchLM public benchmark leaderboard page. source label without registered configuration ID	Qwen3.7-Max	73.6%	CurrentBenchLM aggregated public evaluation	Source carrier · no additional scoring weight source-checked	BenchLM public benchmark leaderboards ↗Observed 2026-07-15 · checked 2026-07-15
Exact model variant as listed on the BenchLM public benchmark leaderboard page. source label without registered configuration ID	Qwen3.6 Plus	71.6%	CurrentBenchLM aggregated public evaluation	Source carrier · no additional scoring weight source-checked	BenchLM public benchmark leaderboards ↗Observed 2026-07-15 · checked 2026-07-15
Exact model variant as listed on the BenchLM public benchmark leaderboard page. source label without registered configuration ID	Qwen3.7-Plus	71.4%	CurrentBenchLM aggregated public evaluation	Source carrier · no additional scoring weight source-checked	BenchLM public benchmark leaderboards ↗Observed 2026-07-15 · checked 2026-07-15
Exact model variant as listed on the BenchLM public benchmark leaderboard page. source label without registered configuration ID	Claude Opus 4.5	70.6%	CurrentBenchLM aggregated public evaluation	Source carrier · no additional scoring weight source-checked	BenchLM public benchmark leaderboards ↗Observed 2026-07-15 · checked 2026-07-15
Exact model variant as listed on the BenchLM public benchmark leaderboard page. source label without registered configuration ID	Kimi K2.5	69.2%	CurrentBenchLM aggregated public evaluation	Source carrier · no additional scoring weight source-checked	BenchLM public benchmark leaderboards ↗Observed 2026-07-15 · checked 2026-07-15
Exact model variant as listed on the BenchLM public benchmark leaderboard page. source label without registered configuration ID	GLM-5	66.8%	CurrentBenchLM aggregated public evaluation	Source carrier · no additional scoring weight source-checked	BenchLM public benchmark leaderboards ↗Observed 2026-07-15 · checked 2026-07-15

Compare top models

Claude Opus 4.6 vs Claude Sonnet 4.6 Claude Sonnet 4.6 vs Qwen3.7-Max Qwen3.7-Max vs Qwen3.6 Plus Qwen3.6 Plus vs Qwen3.7-Plus

About SuperGPQA

Definition and scoring

Organisation: SuperGPQA authors
Category: Reasoning
Version: Not evaluated
Direction: higher is better
Ranking use: Weighted 2.5%
Contamination risk: Unknown

Graduate-level Google-proof Q&A expansion beyond GPQA Diamond for harder knowledge-reasoning items. Every genuine source row stays tied to its exact model label, configuration, benchmark version and evaluation system. The summary chart shows one best compatible score per canonical product; Score 2.0 admits only explicitly mapped, frozen protocols and keeps incompatible configurations separate.

Open benchmark source ↗

FAQ

Quick answers

What does SuperGPQA measure?

Graduate-level Google-proof Q&A expansion beyond GPQA Diamond for harder knowledge-reasoning items.

Which model scores highest on SuperGPQA?

Claude Opus 4.6 by Anthropic currently leads with 95%.

How many models are evaluated on SuperGPQA?

8 canonical model products have a qualifying score, backed by 8 distinct dated source rows.

Does this affect overall Lumina rank?

Yes. This family is ranking-weighted at 2.5% of its capability category. Claude Opus 4.6 is overall #30.

Reasoning · Benchmark profile

SuperGPQA

Graduate-level Google-proof Q&A expansion beyond GPQA Diamond for harder knowledge-reasoning items.

Ranking weightedCurrent / rollingActive8 models

Data verified 15 Jul 2026 · Methodology 2.3.0

Benchmark score on SuperGPQA