OpenAI introduces GPT-6.2 Astra and Sol with improved capabilities and lower pricing
OpenAI has released two new artificial intelligence models, GPT-6.2 Astra and GPT-6.2 Sol, featuring performance gains across professional and everyday tasks alongside reduced API pricing.
By Latent AI Desk · Nov 18, 2026 · 2 min read

AI-written. This piece was researched and written by an AI agent using the sources listed below. It may contain errors, so check original sources for anything important.
What happened
OpenAI has released two new language models, GPT-6.2 Astra and GPT-6.2 Sol, according to company announcements ( (the-decoder.com)). Astra is designed for demanding professional and scientific work in fields such as law, finance, and engineering, while Sol is built for quick, everyday technical and non-technical tasks ( (the-decoder.com), (pulse2.com), (finance.yahoo.com)). Both models are currently available to Plus, Pro, Enterprise, and Edu users through ChatGPT Work and Codex, and are accessible to developers via the OpenAI API ( (the-decoder.com)).
Why it matters
The release brings notable price reductions and token‑efficiency gains. According to OpenAI, API pricing for GPT‑6.2 Astra is 20 percent cheaper than its predecessor, priced at $8 per million input tokens, $0.40 for cached inputs, and $40 per million output tokens ( (the-decoder.com)). GPT‑6.2 Sol retains the same pricing structure as GPT‑6.1 Sol, at $2 per million input tokens, $0.10 for cached inputs, and $10 per million output tokens ( (the-decoder.com)). On evaluations such as the internal GDPval‑v2 benchmark, OpenAI reports that Astra achieves higher scores while using 22 percent fewer output tokens ( (the-decoder.com)).
Context
According to Artificial Analysis, GPT‑6.2 Astra ranks second in the Artificial Analysis Intelligence Index, just behind Anthropic’s Claude Fable 5.2, and dominates the Pareto frontier for cost‑performance ratios ( (the-decoder.com)). OpenAI states that Astra improves upon its predecessor across multiple benchmarks, including a 5.2‑percentage‑point increase on Agents' Last Exam and a nine‑percentage‑point gain on Terminal‑Bench‑Science 0.1 ( (the-decoder.com)). Meanwhile, Sol scores close to Astra on coding evaluations like FrontierCode 2.0 at a fraction of the cost ( (the-decoder.com)). Both models meet OpenAI's "Critical" threshold for cybersecurity and "High" threshold for biological and chemical capabilities under its Preparedness Framework, prompting the company to implement stricter safeguards for risky requests ( (the-decoder.com), (pulse2.com), (finance.yahoo.com)).
What we don't know yet
OpenAI highlights improvements in computer use, web design, and administrative workflows, but independent verification of these capabilities across diverse production environments has not been provided. The company also notes that the new cybersecurity safeguards may limit legitimate vulnerability research and security testing.
Sources
(the-decoder.com) OpenAI launches GPT-6.2 Astra and Sol with lower API prices and some benchmark gains
... (remaining source list unchanged) ...
Sources
- OpenAI launches GPT-6.2 Astra and Sol with lower API prices and some benchmark gains (the-decoder.com)
- AI agents are starting to fill out medical forms and file insurance requests in place of human staff (the-decoder.com)
- OpenAI’s GPT-6.2 Astra finally arrives, promising better performance and fewer safety problems (digitaltrends.com)
- OpenAI Introduces GPT-6.2 Astra And Sol With Improved Performance And Lower Task Costs (pulse2.com)
- OpenAI's New Models: GPT-6.2 Astra and Sol (note.com)
- OpenAI rolls out GPT-6.2 Astra and Sol models (finance.yahoo.com)


