Anthropic releases Claude Sonnet 5.5: 30% faster, lower cost
Original: Anthropic releases Sonnet 5.5, which it calls a significantly cheaper, faster work partner
Why This Matters
Mid-tier model competition is heating up as labs race to win enterprise agentic workloads on cost and speed.
Anthropic on September 28, 2026 released Claude Sonnet 5.5, its mid-tier model, claiming a 30% speed improvement and lower token costs versus Sonnet 5. The model targets everyday tasks like coding and document creation, and is the first Sonnet subject to the same cyber safeguards applied to Opus and Fable.
Anthropic's Sonnet 5.5 arrives roughly three months after Sonnet 5, which was positioned around cost-efficient agentic deployment. The new version shifts the pitch to speed: Anthropic says it is 30% faster than its predecessor with a meaningfully slower token burn rate — a direct appeal to enterprise customers watching API bills. In Anthropic's model lineup, Sonnet sits below Opus in raw capability but above Haiku. That middle-ground position is now its advantage: Anthropic's own benchmarks show Sonnet 5.5 outperforming Opus 5.5 on agentic coding tasks, attributed to its ability to spawn multiple agents without hitting cost ceilings. Perhaps the most notable policy change is on safety. Anthropic says Sonnet 5.5 has 'comparable' cyber capabilities to Opus 5, making it the first Sonnet model brought under the same cyber-specific safeguards previously reserved for Opus and Fable. A new version of Haiku, Anthropic's smallest model, is also on the way, though no release date was given. The release follows OpenAI's recent launch of enhanced versions of Sol and Luna, its own mid-tier and budget models, and a new Meta model tied to its smart glasses.