Vibeleaderboard
← All Intel
Intel / article

GPT-6.2 Astra and Sol land in GitHub Copilot

Source
github.blog
Author
GitHubTop Viber
Date

Context

OpenAI's newest coding models, released on November 17, are now selectable inside GitHub Copilot, but only on its higher paid plans. GitHub's changelog lists Copilot Pro+, Max, Business and Enterprise; the list leaves out the Free and Pro plans. The models appear in the model picker across Copilot's editors, its command-line tool, its coding and github.com, with a gradual rollout.

GitHub positions GPT-6.2 Astra as the model for agentic coding and terminal work, where the AI carries out multi-step changes itself, and GPT-6.2 Sol for complex agentic coding and in-depth code review. Its quality claims come from early testing without published numbers: it says Sol caught nearly twice as many real issues in code review as GPT-6.1 Sol while keeping its findings accurate, and that Astra caught significantly more issues than GPT-6 Astra while using noticeably fewer on coding tasks.

Cost is the practical consideration. Both models are billed at OpenAI's list price under Copilot's usage-based billing. Artificial Analysis reported those list prices as 8 and 40 US dollars per million input and output tokens for Astra, and 2 and 10 dollars for Sol. For companies, the models switch on automatically under Copilot's default model policy unless an administrator has turned that default off or blocked these models, so teams that want to control spending need to check that setting.

Key takeaways · AI-distilled
  • GitHub says that in early testing GPT-6.2 Sol caught nearly twice as many real code-review issues as GPT-6.1 Sol while keeping its findings accurate.
  • GitHub reports that GPT-6.2 Astra caught significantly more code-review issues than GPT-6 Astra and completed agentic coding tasks using noticeably fewer tokens in early testing.
  • On Copilot Business and Enterprise, both models are enabled automatically unless an administrator has turned off the global default or disabled them in model policy, and the rollout is gradual.
Terms in this piece · Glossary
  • AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
  • token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
Read the source github.blog
More from GitHub
Recommended reads
Comments

Checking sign-in…

Loading comments…