Atlas

Models

← All models

GPT-6.1 Sol

OpenAI · 2026-09-29

GPT-6.1 Sol is OpenAI's September 29, 2026 reasoning model for coding, computer use, and professional work. It accepts text and image inputs and produces text, with low, medium, high, xhigh, and max reasoning effort. Tool calling uses the Responses API. Its documented total context window is 1,050,000 tokens, with at most 922,000 input tokens and 128,000 output tokens.

Benchmark scores

BenchmarkScore
AA-3D v1.0 Elo1119
AA-AnalystAgent (pass@1)59.8
AA-AnalystAgent (pass@5)72.5
AA-AnalystAgent (pass^5)50.0
AA-Briefcase v1.1 Elo1555
AA-LCR v1.184.0
AA-Omniscience Index41.5
AA-Omniscience v1.1 Accuracy61.8
AA-Omniscience v1.1 Hallucination Rate48.0
AA-Omniscience v1.1 Index42.4
AA-Robotics v0.164.6
ActiveVision71.8
AdvancedIF82.4
AdvancedIF (rubric mean score)94.8
Agent Red Teaming k=100 (CAIS)18.8
AIM Consistency98.4
AIM Grounding94.2
AIM Routing89.3
AIME 2025100.0
ALE-Bench2990
APEX (Mercor)20.1
APEX-Agents 1.173.2
APEX-Agents 1.1 Corporate Law (68-task image-free slice)85.3
APEX-Agents 1.1 Pass@160.0
APEX-SWE46.9
APEX-v1-extended69.3
ARC-AGI-1 Semi-Private98.5
ARC-AGI-294.2
ARC-AGI-3 (Semi-Private)96.4
Arena Alignment — Deceptive Completion2.3
Loading Atlas data…