GitHub: Model Selection for Claude and Codex Agents Now Available
GitHub now allows developers to choose between multiple AI models when launching Claude and Codex coding agents. Available models include Claude Sonnet/Opus 4.5 and 4.6 as well as GPT-5.2/5.3/5.4-Codex.
This article was generated using artificial intelligence from primary sources.
On April 14, GitHub announced a new feature that allows developers to select AI models when launching coding agents directly on github.com.
Available Models
Claude agent (Anthropic):
- Claude Sonnet 4.6 and Opus 4.6 (latest)
- Claude Sonnet 4.5 and Opus 4.5
Codex agent (OpenAI):
- GPT-5.2-Codex, GPT-5.3-Codex, and GPT-5.4
Users can select a model when launching each task, similar to how the Copilot cloud agent already works.
Access and Pricing
Access to Claude and Codex agents is included in the existing Copilot subscription — no additional costs. However, for Copilot Business and Enterprise users, an administrator must approve the relevant policies, and the repository owner must enable agents in the settings.
Why This Matters
This is a significant step toward a model-agnostic approach to AI in development tools. Developers are no longer tied to a single model — they can choose the optimal model for each specific task. Complex refactoring can use Opus 4.6, while routine tasks can run on the faster Sonnet 4.6.
Related news
arXiv:2605.22681: CUSP benchmark shows frontier models cannot reliably predict scientific breakthroughs
arXiv:2605.22337: Meta-Soft introduces KV cache compression via composable meta-tokens and learnable orthogonal bases
arXiv:2605.22664: WorkstreamBench tests LLM agents on end-to-end spreadsheet tasks in finance — and frontier models fail