OpenAI has introduced GPT-6.1 Sol, a cost-effective model that offers nearly comparable capabilities to the high-performance GPT-6 Astra for professional work and coding, priced at a fraction of Astra's token costs.
- GPT-6.1 Sol costs a fifth of GPT-6 Astra’s token rates
- Matches Astra on coding and business workflow tests
- Planned Ultrafast tier will boost generation speeds next
What happened
OpenAI unveiled GPT-6.1 Sol during DevDay 2026, positioning it as a more affordable alternative to its GPT-6 Astra model. It costs $2 per million input tokens and $10 per million output tokens, significantly cheaper than Astra and previous Sol versions. The model supports agentic coding, computer tasks, and professional workflows and is immediately accessible to Plus, Pro, Business, Enterprise, and Edu users via ChatGPT Work and Codex, as well as through the API under the identifier gpt-6.1-sol.
Alongside price reductions, OpenAI announced plans to launch an Ultrafast tier designed to produce tokens up to eight times faster in Codex, enhancing developer experience. OpenAI also provided performance metrics illustrating Sol's competitive accuracy and safety performance relative to Astra and competitors, acknowledging ongoing safety evaluations and model improvements following past cancellations of other model releases.
Why it matters
The release of GPT-6.1 Sol marks a substantial shift in AI pricing dynamics, enabling wider adoption of advanced generative models by lowering token usage costs dramatically. This affordability can accelerate integration of AI into software engineering, business automation, and scientific research contexts, previously limited by high operational costs.
Sol's competitive performance in key benchmarks—including software engineering tests (DeepSWE), business workflow automation (AutomationBench), and computer-use evaluations (OSWorld)—demonstrates that cost savings do not necessarily mean a sacrifice in capability. Customer feedback, such as from Salesforce's EVP of software engineering, validates Sol's practical utility in identifying real-world software issues and improving testing comprehensiveness.
What to watch next
OpenAI’s forthcoming Ultrafast tier implementation for GPT-6.1 Sol will be critical to monitor, as it promises to greatly enhance response times and throughput for developers using the Codex environment. This could further differentiate Sol as a versatile model for demanding coding and business applications.
Additionally, ongoing safety assessments and refinement are important to track, particularly following the recent cancellation of GPT-6.1 Astra's launch due to failed safety tests. OpenAI’s ability to balance cutting-edge model performance with robust safeguards will influence market trust and adoption rates. Industry watchers should also observe how competitors respond in pricing and capabilities as OpenAI reshapes the AI model marketplace.