OpenAIreleased 22 Sep 2026

GPT-6 Sol

#11 of 9527 of 46 benchmarksConsistent

84.7

B.O.B.

Category profile

Real use (hard to game)2×79
Reasoning & knowledge84
Math91
Coding87
Agents & real work82
Independent indexes91

Reality check

Lab tests

Real use (7 tests)

Lab tests and real-use tests agree. It does 7 points better on the usual lab tests than on tests that are hard to train for.

What counts as real use

Price per 1M tokens (OpenRouter)

$2.00 in · $10.00 out

Cost per task on the AA index

—

Output speed (Artificial Analysis)

—

§Receipts

Every benchmark result

“Beats” = the share of tracked models this one beat on that test. The raw score is its best reported setting.

Benchmark	Raw	Beats	Source
Real use (hard to game) · Real use 79
EQ-Bench Creative Writingas gpt-6-sol	2125	93%of 46	Sam Paech (open source)
LMArena WebDevas gpt-6-sol-max	1686	93%of 82	LMArena (blind human votes)
LMArena Agent Arenaas GPT 6 Sol (Max)	0.105	88%of 50	LMArena (real agent sessions)
BridgeBench (overall)as GPT-6 Sol	643	82%of 29	BridgeMind
LiveBench 2.0 (real-work suites)as gpt-6-sol	80.7%	79%of 39	LiveBench / Abacus.AI
Vals Indexas openai/gpt-6-sol	57.5%	73%of 42	Vals AI (private test sets)
LMArena Textas gpt-6-sol-max	1458	47%of 80	LMArena (blind human votes)
Reasoning & knowledge · Reasoning 84
CritPt (physics research)as gpt-6-sol_max	30.9%	93%of 89	Epoch AI Benchmarking Hub
GPQA Diamondas gpt-6-sol_max	94.3%	88%of 70	Epoch AI Benchmarking Hub
ARC-AGI-2as gpt-6-sol_max	89.6%	86%of 44	Epoch AI Benchmarking Hub
Mystery Game Puzzlesas gpt-6-sol_max	56.0%	85%of 57	Epoch AI Benchmarking Hub
Chess Puzzlesas gpt-6-sol_max	56.0%	85%of 73	Epoch AI Benchmarking Hub
SimpleQA Verifiedas gpt-6-sol_max	60.7%	77%of 57	Epoch AI Benchmarking Hub
Blueprint-Bench 2as gpt-6-sol_unknown	36.9%	77%of 31	Epoch AI Benchmarking Hub
Math · Math 91
OTIS Mock AIMEas gpt-6-sol_max	100.0%	94%of 73	Epoch AI Benchmarking Hub
FrontierMath Tiers 1-3as gpt-6-sol_max	89.8%	93%of 60	Epoch AI Benchmarking Hub
FrontierMath Tier 4as gpt-6-sol_max	90.0%	92%of 52	Epoch AI Benchmarking Hub
ProofBenchas gpt-6-sol_unknown	83.0%	83%of 68	Epoch AI Benchmarking Hub
Coding · Coding 87
ALE-Benchas gpt-6-sol_max	2462.0%	97%of 71	Epoch AI Benchmarking Hub
FrontierCodeas gpt-6-sol_max	49.3%	82%of 45	Epoch AI Benchmarking Hub
SciCodeas gpt-6-sol_max	57.6%	81%of 86	Epoch AI Benchmarking Hub
Agents & real work · Agents 82
Vending-Bench 2as gpt-6-sol_unknown	14427.8%	98%of 47	Epoch AI Benchmarking Hub
DTBenchas gpt-6-sol_max	97.3%	92%of 67	Epoch AI Benchmarking Hub
LMCAas gpt-6-sol_max	59.1%	91%of 67	Epoch AI Benchmarking Hub
GDP.pdfas gpt-6-sol_max	26.4%	72%of 37	Epoch AI Benchmarking Hub
APEX-Agentsas gpt-6-sol_max	54.3%	55%of 48	Epoch AI Benchmarking Hub
Independent indexes · Indexes 91
Epoch Capabilities Indexas GPT-6 Sol	162.6	91%of 81	Epoch AI