Anthropic released Claude Sonnet 5.5 across its API, apps, Claude Code, AWS, Google Cloud and Microsoft Azure. Pricing stays at $2 per million input tokens and $10 per million output tokens, but Anthropic says the model generates 30% faster and costs up to 30% less per task because it uses fewer tokens. Vendor evaluations put it close to Opus 5.5 on several coding and knowledge-work tests, with a 1M context window. The launch also changes safeguard behavior: some high-risk cyber requests fall back to Sonnet 5, and preserved thinking now binds reasoning to the account that created it.
Anthropic releases Claude Sonnet 5.5 with faster, cheaper agent work
2026-09-29Verified announcementTRIAL
Claude Sonnet 5.5 is broadly available at $2 per million input tokens and $10 per million output tokens. Anthropic reports faster, lower-cost agent work, but Artificial Analysis found that maximum effort used about 193k output tokens per task and cost roughly $7.60, around 50% more than Sonnet 5 in its suite. Cost savings therefore depend on effort and workload.
Official · AnnouncementAnthropic Claude Sonnet 5.5 announcementOfficial · Release notesAnthropic Claude Code 2.1.284 release
Availability: Claude Platform, Claude apps, Claude Code, AWS, Google Cloud and Microsoft Azure
What this means for you
For model-routing teams, the vendor cost claim should not be treated as a universal discount. Benchmark low, medium and maximum effort on your own traces and route by quality-adjusted task cost.
Evidence & related material
Independent evaluation finds Sonnet 5.5 maximum effort can cost more
Subsequent changes
Independent evaluation finds Sonnet 5.5 maximum effort can cost more
Technical details & scores
| Model | claude-sonnet-5-5 | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| Context Window | 1M tokens | ||||||||
| Pricing Per Million Tokens |
| ||||||||
| Vendor Speed Claim | 30%+ faster generation than Sonnet 5 | ||||||||
| Vendor Cost Claim | up to 30% lower cost per task than Sonnet 5 | ||||||||
| Vendor Benchmarks |
| ||||||||
| Availability | Claude Platform, Claude apps, Claude Code, AWS, Google Cloud and Microsoft Azure | ||||||||
| Migration | thinking-off clients must use between_tools before switching | ||||||||
| Caveat | performance, speed and per-task cost claims are vendor-reported; some structured-output benchmark runs used a prerelease deployment with a later-fixed bug |
Technical impact 4
Engineering value 5
Adoption signal 4
Maturity 4
Verification cost 2
Risk 3