On November 17, 2026, OpenAI released GPT-6.2 Astra and Sol, and the rollout says as much about where OpenAI thinks its models belong as the models themselves. Both are available from today in ChatGPT Work and Codex for Plus, Pro, Enterprise and Edu users, and to developers through the API as gpt-6.2-astra and gpt-6.2-sol. They are not in regular Chat, and free accounts do not get them. Astra is the frontier model for demanding projects that need reasoning across code, applications or research, while Sol is the less expensive model for everyday coding and writing. OpenAI's pitch for both is the same: better results on demanding tasks with fewer output tokens, so you get more done within your usage limits. For the first time, Ultrafast mode comes to Pro 200 subscribers, starting with GPT-6.2 Sol, and the same day also brought ChatGPT for Sales, which writes CRM updates among its functions.
Prices, context, and the numbers OpenAI reports
In the API, Astra costs $8 per million input tokens and $40 per million output tokens, with cached input at $0.40, which is 20 percent below GPT-6 Astra. Sol keeps GPT-6.1 Sol's prices at $2 input and $10 output, with cached input at $0.10. For requests above 272,000 input tokens, both models double the input and cache rates and charge 1.5 times the output rate for the whole request. Both take text and images in and return text, with a 1,050,000-token context window, up to 922,000 input tokens and up to 128,000 output tokens. The knowledge cutoffs are May 18, 2026 for Astra and June 1, 2026 for Sol. On OpenAI's own benchmarks, Astra jumps 22 points over GPT-6 Astra on ExploitBench Internal Port, and Sol comes within half a point of Astra at maximum reasoning effort on Cognition's FrontierCode 2.0 for about one fifth the cost. Independent evaluator Artificial Analysis puts Astra at 62 on its Intelligence Index, up from 57 for GPT-6 Astra, and Sol at 57, up from 55 for GPT-6.1 Sol, with a weighted cost per index task of $4.00 for Astra and $1.01 for Sol. One price worth noting: Astra now costs the same per token as Anthropic's Claude Fable 5.2, which Anthropic cut by 20 percent a week earlier. OpenAI also says both models reached a critical threshold for cybersecurity capabilities, so they ship with tighter safeguards for higher-risk cybersecurity assistance.
The migration change that touches your code
Changing the model string is the easy part of this upgrade. Starting with GPT-6.2, the API drops earlier reasoning items when it detects edits to earlier conversation history, a model downgrade, a changed end user or a changed organization, so any app that rewrites history between turns needs to handle that. Tool calling works only through the Responses API, not Chat Completions, and neither model accepts a reasoning effort setting. The API also adds 2-hour and 8-hour background completion windows at a discount. Ultrafast is priced at 3 times Standard through February 28, 2027 and at 6 times Standard from March 1, 2027. OpenAI also says that from December, Plus, Pro and Business plans will share one allowance across ChatGPT and Codex, measured in tokens, so the per-token efficiency of these models is what decides how much work a plan covers. Before you switch, pin the model ID, re-run your evals, and check the limits page for your plan, because availability depends on plan, workspace permissions and rollout progress. Our GPT-6.1 Terra explainer covers the previous generation these prices are measured against, and our Claude Fable 5.2 explainer covers the Anthropic model that now costs the same per token as Astra. If you want the model choice, the spend alerts, and the monitoring around an AI workload checked by someone who runs infrastructure for a living, book a free infra audit with DevOpsBY.
Disclaimer: This post is educational only, not financial advice. Figures are as reported by the sources on 2026-11-17 and may change - always confirm version numbers, requirements, and prices in the official docs before acting on them.
