Claude Sonnet 5 is Anthropic’s current Sonnet-class model: a commercial model for coding, agents and professional knowledge work. It is available through the Claude API and selected cloud platforms, and uses adaptive thinking by default.
Its 1 million-token context window and 128,000-token maximum output remain in place. The $2 / $10 per million input/output rates announced as introductory pricing through 31 August 2026 are now the standard price; Anthropic did not apply the previously scheduled move to $3 / $15.
At a glance
| Fact | Detail |
| --- | --- |
| Provider | Anthropic |
| Type | Commercial / proprietary |
| Best for | Coding, agents, automation, professional work |
| Context window | 1,000,000 tokens |
| Maximum output | 128,000 tokens |
| API available | Yes |
| Local use | No |
| Last checked | 31 August 2026 |
Our verdict
Claude Sonnet 5 remains one of the strongest practical choices for coding and agentic workflows when you want high capability without Opus-class pricing. Keeping $2 / $10 as the standard rate, rather than stepping up on 1 September, makes the cost case clearer. Tokenizer differences versus Sonnet 4.6 still mean token-for-token comparisons need care.
| Area | Score |
| --- | --- |
| Overall | ★★★★☆ |
| Coding | ★★★★★ |
| Agents | ★★★★★ |
| Research | ★★★★☆ |
| Budget | ★★★★☆ |
Pricing
Standard Claude API prices per million tokens:
| Meter | Price |
| --- | --- |
| Input | $2.00 / million tokens |
| Cached input | $0.20 / million tokens |
| Output | $10.00 / million tokens |
Anthropic’s live pricing table lists these as the current Sonnet 5 rates. Batch processing and prompt-cache writes can differ; check the official table for those meters.
Best for
- Coding, agents and professional knowledge work
- Repository-scale context with adaptive thinking
- Teams that want Claude quality below Opus spend
Avoid when
- You need local inference or open weights
- The job needs Opus- or Fable-class capability
- You cannot tolerate a hosted-provider dependency
What changed
- 31 August 2026: Official pricing re-checked. Anthropic states the $2 / $10 introductory rates are now the standard price; the scheduled $3 / $15 increase was not applied.
- 16 July 2026: Profile created; pricing, context and API availability checked against Anthropic documentation.
- 30 June 2026: Anthropic launched Claude Sonnet 5 (
claude-sonnet-5) with 1M context, 128K maximum output and adaptive thinking enabled by default.
Related news
- Alibaba Bans Employees from Using Anthropic’s Claude Code
- SpaceXAI Ships Grok 4.5, Musk Calls It an Opus-Class Model at Half the Price
- OpenAI’s GPT-5.6 Push Is Really a Pricing War for Enterprise AI
