Claude Sonnet 5
Last updated: August 11, 2026
Specs
| Vendor | Anthropic |
|---|
| Released | June 30, 2026 |
|---|
| Context window | 1M tokens |
|---|
| Input | $2 / M tokens |
|---|
| Output | $10 / M tokens |
|---|
Benchmarks
Strengths
- Frontier-class coding and agent work at Sonnet pricing
- 1M-token context
- Adaptive effort, so cost tracks task difficulty
Weaknesses
- Opus still leads on the very hardest tasks
- Adaptive effort makes per-call cost harder to predict
Best for
- Production coding copilots
- Agents that must justify their cost
- Most MVP and SaaS workloads
Compare with other AI models