GPT-6.1 Sol

2026-09-29 · Publisher : OpenAI · Domain : Artificial Intelligence

OpenAI launches GPT-6.1 Sol, an upgrade to GPT-6 Sol that approaches GPT-6 Astra in agentic coding, computer use and professional work, at one-fifth of Astra's standard API prices.

### In detail - Sol gets closer to Astra on demanding tasks, while Astra remains the most capable model in the lineup. - Cached input at $0.10 per million tokens, 95% less than standard input: agents that reuse their context cost less. - Improved alignment over GPT-6 Sol, more transparent about its limits and more faithful to explicit constraints. ### Specifications - Context: 1,050,000 tokens (922,000 input maximum), maximum output 128,000 tokens, text and image input, text output. - Knowledge cutoff: April 30, 2026. Effort: low, medium (default), high, xhigh, max; no none or minimal. - Tool calling through the Responses API; Chat Completions without tools. US and EU data residency. - Availability: ChatGPT Work and Codex (Plus, Pro, Business, Enterprise, Edu), API as gpt-6.1-sol; not yet in Chat. | Input | Cached input | Cache write | Output | | --- | --- | --- | --- | | $2.00 | $0.10 | $2.50 | $10.00 | Price per million tokens (USD). Beyond 272,000 tokens: 2x input, 1.5x output. Vendor-published figures: DeepSWE v1.1 on par with Astra for about one-fifth of the cost; OSWorld 2.0 within 2.1 points of Astra; $5.47 per task on Terminal-Bench Science 0.1 versus $23.80 for Astra. ### Why it matters The best performance-to-cost ratio in OpenAI's lineup becomes the default candidate for high-volume coding and automation agents. ### Key takeaways The GPT-6.1 Sol Ultrafast variant is announced but not yet priced in the API. Compare it with Astra on your own tasks.

Tags : LLM, OpenAI, GPT-6, Sol, Agent, Computer Use