Claude Sonnet 5.5

Claude Sonnet 5.5 is a large language model developed by Anthropic and released on 28 September 2026 as the second model in the Claude 5.5 family, following Claude Opus 5.5. Anthropic positions Sonnet 5.5 as a faster, cheaper complement to Opus 5.5, strongest at well-scoped everyday tasks — bug fixes, documents, slides and spreadsheets — while retaining near-flagship coding ability.^[1]

History

Anthropic introduced the Claude 5.5 generation with Opus 5.5 on 22 September 2026, describing it as its strongest model to date, then released Sonnet 5.5 six days later. Sonnet 5.5 succeeds Claude Sonnet 5, released on 30 June 2026, and is described by Anthropic as "a clear upgrade over Claude Sonnet 5" that runs 30%+ faster and costs up to 30% less for most work. A third family member, Claude Haiku 5.5, was announced for high-volume, cost-sensitive applications.^[1][2]

Key concepts and technology

Sonnet 5.5's headline improvement is agentic coding: it scores 70.6% on Terminal-Bench 4.0, compared with 10.3% for Sonnet 5, and 55.5% on CursorBench — second only to Opus 5.5 among Claude models. Anthropic reports it sits two points below Opus 5.5 on GDPval-AA, a benchmark of real-world occupational work, and that it is the first Sonnet model to complete Pokémon Red working only from screenshots — a common test of long-horizon planning and visual reasoning. Early testers described clearer writing and faster iteration than previous generations.^[1]

Pricing is unchanged from Sonnet 5 at $2 per million input tokens, $10 per million output tokens and $0.20 per million cached tokens, but better token efficiency means typical tasks cost up to 30% less, and outputs are generated 30%+ faster than Sonnet 5.^[1][3]

On safety, Sonnet 5.5 is the first Sonnet model shipped with cyber safeguards patterned on those used for Anthropic's most capable models, because its cybersecurity capabilities approach Opus-class levels; its biological safeguards match Sonnet 5's. Anthropic states the safeguards target a narrow set of high-risk requests, leaving routine software development and most life-science work unaffected. Availability includes claude.ai and the Claude API under the model identifier "claude-sonnet-5-5", plus Amazon Web Services, Google Cloud and Microsoft Foundry, with zero data retention available.^[1][3]

Applications and impact

Sonnet 5.5 targets the volume tier of professional AI use: software teams running coding agents, financial and legal analysts processing documents, and customer-facing automation where latency matters. Anthropic's own evaluations and early customers — including GitHub, Spotify and Lovable — reported completing the same work in fewer steps and with fewer tokens than Sonnet 5. Its cost and speed position make it the default choice where Opus 5.5's capability would exceed the task's requirements.

>Key Takeaways

• Released 28 September 2026 as the second member of the Claude 5.5 family after Opus 5.5.
• Scores 70.6% on Terminal-Bench 4.0 versus 10.3% for Sonnet 5; runs 30%+ faster and costs up to 30% less per task.
• First Sonnet model with cyber safeguards; priced at $2 input / $10 output per million tokens.

See Also

• Claude Language Model
• Anthropic
• Reasoning Models
• Code Generation
• Agentic AI