OpenAI officially launched its new GPT-6.1 Sol artificial intelligence model on September 29, 2026. The release rolls out the model across GitHub Copilot and select ChatGPT enterprise tiers to significantly enhance agentic coding and multi-step computational workflows.
The updated model delivers intelligence nearly on par with the flagship GPT-6 Astra for complex programming, intensive document processing, and computer use tasks. Crucially, it achieves this high performance while operating at just one-fifth of the standard input and output token prices of its predecessor.
In early performance testing, the system completed benchmark tasks using noticeably fewer steps and tokens compared to prior models in the GPT-6 and GPT-5.6 lineups. This operational streamlining makes it an attractive option for developers utilizing GitHub Copilot for large-scale software engineering.
Measurable Reductions in Hallucination Risks
Alongside significant pricing efficiencies, the update demonstrates measurable improvements in reliability and safety. On difficult prompts evaluated under low reasoning settings, the percentage of responses containing factual errors fell from 11.4 percent in standard versions down to 7.7 percent in GPT-6.1 Sol.
, the new model maintains a tight error rate that stays within 1.9 percent of the more resource-intensive GPT-6 Astra across all tested reasoning configurations.
Expanded Impact on Enterprise Workflows
Industry analysts note that the introduction of GPT-6.1 Sol bridges the gap between cost-effective execution and top-tier computational reasoning. By lowering token costs while improving accuracy on complex software tasks, OpenAI aims to accelerate enterprise adoption of autonomous coding agents.
Key Highlights of GPT-6.1 Sol
- Release Date: September 29, 2026
- Platform Availability: GitHub Copilot and select ChatGPT enterprise tiers
- Cost Efficiency: Operates at one-fifth of standard input and output token prices
- Error Reduction: Factual error rates dropped to 7.7 percent under low reasoning settings
