Sonnet family · Released June 30, 2026
Available nowClaude Sonnet 5
Anthropic's most agentic Sonnet model — planning, tool use, and autonomous coding at close to Opus quality, for a fraction of the price. Released June 30, 2026 as claude-sonnet-5.
- $2 / $10
- per M in / out (intro, through Aug 31)
- 1M
- token context window
- Jun 30, 2026
- launch date
- claude-sonnet-5
- API model ID
Why Claude Sonnet 5 matters
- Most agentic Sonnet yet — it plans multi-step work, drives browsers and terminals, and runs autonomously at a level that recently required larger, pricier models.
- 1M-token context window by default with up to 128K output tokens — long codebases and documents fit in a single call.
- The new default model for Free and Pro plans, and available to Max, Team, Enterprise, Claude Code, and the Claude API on day one.
- Introductory pricing of $2 / $10 per million tokens through August 31, 2026 — well under half the cost of Opus 4.8 at comparable quality.
- Adaptive thinking is on by default — the model decides how much to reason per request, so you get depth on hard problems without paying for it on easy ones.
- Drop-in upgrade from Sonnet 4.6: same tools and platform features, just update the model ID to claude-sonnet-5 (manual extended thinking and non-default sampling params now return 400).
Claude Sonnet 5 specifications
| — | Claude Sonnet 5 |
|---|---|
| API model ID | claude-sonnet-5 |
| Release date | June 30, 2026 |
| Context window | 1,000,000 tokens |
| Max output | 128K tokens |
| Pricing (in / out per M) | $2 / $10 (intro) → $3 / $15 |
| Thinking | Adaptive by default — manual extended thinking and non-default sampling params return 400 |
| Availability | Claude apps, Claude Code, API, AWS Bedrock, Google Vertex, Microsoft Foundry, Cursor, VS Code, GitHub Copilot, OpenRouter |
Sonnet 5 uses the tokenizer introduced with Claude Opus 4.7, which produces roughly 30% more tokens for the same text than models released before it. Per-task cost is therefore higher than Sonnet 4.6 at an unchanged list price — but identical in tokenization to Fable 5 and Opus 5, which use the same tokenizer. Sources: Anthropic, June 30, 2026 and the Claude token-counting documentation.
Claude Sonnet 5 vs Claude Fable 5
Fable 5 is Anthropic's Mythos-class model for the hardest, longest work at $10/$50; Sonnet 5 is the fast, agentic mid-tier at $2/$10. Choose Sonnet 5 for high-volume coding and tool use, Fable 5 when its peak reasoning quality justifies the cost — and compare against Opus 5 at $5/$25 before you do.
Read the full Sonnet 5 vs Fable 5 guideClaude Sonnet 5 FAQ
Build on Claude Fable5 with your eyes open
Know the Claude Fable5 cost math, the safeguard behavior, and the right tier before you commit. Create a free account to get notified when specs, prices, or access channels change.