On September 29, 2026, OpenAI held DevDay 2026, its annual developer event. The headline was the new GPT-6.1 Sol model, alongside a string of announcements on agent development and Codex.
GPT-6.1 Sol: near top-tier performance at one-fifth the price
GPT-6.1 Sol comes close to OpenAI's top-tier GPT-6 Astra in agentic coding, computer use and professional work, while its standard price is one-fifth of Astra's.
- API pricing of $2 per million input tokens and $10 per million output tokens
- Cached input costs $0.10, half of GPT-6 Sol's $0.20
- 75.2% on the DeepSWE v1.1 coding evaluation (High reasoning effort), beating GPT-6 Sol at its maximum setting (68.8%)
- 71.4% on the offline set of the OSWorld 2.0 computer-use evaluation (Astra: 73.5%; both at maximum reasoning effort), at roughly one-seventh of Astra's cost per task
OpenAI recommends Astra when you want the highest possible quality, and GPT-6.1 Sol for complex work you need to run repeatedly.
GPT-6.1 Sol is available in ChatGPT Work, in Codex, and through the API (model name: gpt-6.1-sol).
Other major announcements
- Codex Cloud: pick up a task from another device even after you close your laptop
- Agents API: a public beta with a hosted runtime, memory and multi-agent support
- Computer use: agents can operate software through on-screen actions
- Dots: always-on agents that connect to apps and keep running in the cloud
- ChatGPT Space: a shared space where teams and agents work together
- Ultrafast mode: a paid option that significantly boosts generation speed
- Sign in with ChatGPT: lets people log in to other services with their ChatGPT account
On the infrastructure side, OpenAI says time to first token in the API is 45% shorter and tool calls are more than 30% faster.
CoAI's take
GPT-6.1 Sol brings a high-performance model closer to a price suited for everyday use. If you use ChatGPT or Codex for work, choosing a model to match each task makes it easier to balance cost and quality. And with the Agents API and Dots, ChatGPT is expanding from a tool you simply talk to into an agent you can hand work off to.
