GPT-6 Sol
The GPT-6 workhorse for complex coding and agentic workflows — Astra-level reliability at a fraction of Astra's price
GPT-6 Sol
OpenAI • September 2026
Training Data
Not disclosed
GPT-6 Sol
September 2026
Parameters
Not disclosed
Training Method
Reinforcement Learning with Chain-of-Thought Reasoning
Context Window
1,050,000 tokens
Knowledge Cutoff
April 20, 2026
Key Features
1M Token Context • Reasoning Effort: None to Max • Agentic Coding • Half the Mistakes of GPT-5.6 Sol
Capabilities
Coding: Maximum (68.8% DeepSWE v1.1)
Agentic Work: Outstanding (56.4% Agents' Last Exam)
Computer Use: Outstanding (60.5% OSWorld 2.0 offline)
What's New in This Version
Makes about half as many mistakes as GPT-5.6 Sol, reaching Astra-level reliability at much lower cost; edges Claude Opus 5 on OSWorld 2.0 offline (60.5% vs 60.3%) at ~80% lower cost per task; $2/$10 per 1M tokens
The GPT-6 workhorse for complex coding and agentic workflows — Astra-level reliability at a fraction of Astra's price
What's New in This Version
Makes about half as many mistakes as GPT-5.6 Sol, reaching Astra-level reliability at much lower cost; edges Claude Opus 5 on OSWorld 2.0 offline (60.5% vs 60.3%) at ~80% lower cost per task; $2/$10 per 1M tokens
Technical Specifications
Key Features
Capabilities
Other OpenAI Models
Explore more models from OpenAI
GPT-6.1 Sol
An upgrade to GPT-6 Sol that nearly matches GPT-6 Astra on agentic coding, computer use, and professional work at one-fifth of Astra's token prices
GPT-6 Luna
OpenAI's most efficient model for focused, high-volume tasks — summarizing documents, extracting information, and answering quick questions
GPT-6 Astra
OpenAI's GPT-6 flagship — a generational leap in computer use, software engineering, science, and professional work that reaches the Critical cybersecurity threshold under OpenAI's Preparedness Framework
