Models / GPT-6.1 Sol

GPT-6.1 Sol

OpenAI · Closed weights · Released Sep 29, 2026 · Price not listed

80Overall · rank 5

See GPT-6.1 Sol alternatives

Post toX LinkedIn Bluesky Reddit Hacker News

Score by job

~ is an early read: this model has results on too few benchmarks in that job to be ranked yet.

Compare GPT-6.1 Sol with

Badge for your README

Shows this model’s current rank. It updates with every data update.

Badge job

Markdown

[![GPT-6.1 Sol, #2 in Reasoning on Benchfolio, score 92](https://benchfolio.pages.dev/badge/gpt-6-1-sol--reasoning.svg)](https://benchfolio.pages.dev/model/gpt-6-1-sol)

Coding

Early read ~71 · not ranked · 2 of 5 benchmarks

Benchmark	Result	Variant	z
FrontierCode	50.2%	gpt-6.1-sol_medium	0.91
LMArena Coding	1487 rating	gpt-6.1-sol-max	-0.10

Agents

Early read ~69 · not ranked · 1 of 4 benchmarks

Benchmark	Result	Variant	z
APEX-Agents	60.0%	gpt-6.1-sol_max	0.65

Reasoning

92 · rank 2 · 2 of 4 benchmarks

Benchmark	Result	Variant	z
GPQA Diamond	95.4%	gpt-6.1-sol_max	1.49
ARC-AGI-2	94.2%	gpt-6.1-sol_max	1.27

Math

91 · rank 6 · 5 of 5 benchmarks

Benchmark	Result	Variant	z
FrontierMath Tiers 1–3	93.7%	gpt-6.1-sol_max	1.60
FrontierMath Tier 4	100.0%	gpt-6.1-sol_max	2.09
OTIS Mock AIME	100.0%	gpt-6.1-sol_max	0.95
ProofBench	99.0%	gpt-6.1-sol_unknown	1.71
LMArena Math	1464 rating	gpt-6.1-sol-max	-0.44

Writing

50 · rank 29 · 2 of 2 benchmarks

Benchmark	Result	Variant	z
LMArena Creative Writing	1432 rating	gpt-6.1-sol-max	-0.41
LMArena Instruction Following	1468 rating	gpt-6.1-sol-max	0.13

Design

87 · rank 4 · 1 of 1 benchmarks

Benchmark	Result	Variant	z
LMArena WebDev	1755 rating	gpt-6.1-sol-max	1.78