Claude Sonnet 5 is now generally available in GitHub Copilot, announced in GitHub's changelog on 2026-06-30. It is Anthropic's newest Sonnet-class model, positioned for everyday development and agentic workflows, and it works across the IDE and the CLI on paid Copilot plans.
What Claude Sonnet 5 is
Claude Sonnet 5 is the latest model in Anthropic's Sonnet line, the successor to Claude Sonnet 4.5. "Generally available" (GA) means it is out of preview and ready for day-to-day use, not a limited test. The Sonnet class is Anthropic's balanced tier, built for the kind of routine coding most developers do all day rather than positioned as a heavyweight reserved for the hardest problems.
This is a different kind of announcement from the Opus 4.8 fast mode preview that landed the same day. That one is about speed: a faster variant of an existing model. This one is about a new model reaching general availability. If you have been waiting for Sonnet 5 to leave preview before relying on it, that wait is over.
Who gets it
Per GitHub's changelog, Claude Sonnet 5 is available to:
- Copilot Pro
- Copilot Pro+
- Copilot Max
- Copilot Business
- Copilot Enterprise
That spread is worth noting: it reaches Pro, the entry paid tier, not only the premium plans. You select it from the Copilot model picker in your IDE or CLI.
What it is good at
In Anthropic and GitHub internal testing, Sonnet 5 showed strong results across a range of coding scenarios, with a few standout traits:
- CLI-style tasks: it performed particularly strongly on command-line work, which matters as more development moves into terminal-driven and agent-run flows.
- Prompt-cache utilization: it demonstrated excellent use of prompt caching, which helps keep repeated, context-heavy interactions efficient.
- Latency at lower effort levels: it showed competitive latency when run at lower effort settings, useful for quick, interactive work.
These are qualitative findings from internal testing, not published benchmark scores. The practical read: Sonnet 5 is aimed squarely at the interactive, agentic, terminal-heavy way many developers now work, rather than at one-off heavyweight reasoning.
How it fits alongside the other Claude options
Copilot now offers several Claude choices, so the real question is which one for what:
- Claude Sonnet 5: a solid default for everyday development and agentic workflows, now GA across IDE and CLI.
- Claude Opus 4.8 (and its fast mode preview): the heavier Opus class, with a speed-optimized variant for latency-sensitive work.
If you are mapping out which assistant fits your stack more broadly, see our best AI coding assistants 2026 roundup and GitHub Copilot alternatives.
How billing works
Claude Sonnet 5 is billed at provider list pricing under Usage-Based Billing (UBB). In practice that means usage is metered and charged at the provider's list rate rather than bundled into a flat allowance. GitHub did not publish a specific per-token figure in the changelog, so check your Copilot billing settings for the current rate before leaning on it heavily.
How to enable it
- Make sure you are on a supported plan (Pro, Pro+, Max, Business, or Enterprise).
- Open the Copilot model picker in your IDE or CLI.
- Select Claude Sonnet 5.
- Use it for everyday edits and agent runs, and check Usage-Based Billing settings to keep an eye on usage.
The bottom line
Claude Sonnet 5 reaching general availability in GitHub Copilot gives developers on every paid tier, from Pro to Enterprise, a current Sonnet-class model built for everyday and agentic coding across the IDE and CLI. Its internal-testing strengths, CLI tasks, prompt caching, and low-effort latency, line up with terminal-driven, agent-run workflows. Billed under UBB at provider list pricing, it is a sensible default to reach for, with Opus on hand when you want the heavier class.
For broader comparisons, read Cursor vs Claude Code and our best AI coding assistants 2026 overview.
