Anthropic ships Claude Sonnet 5.5, matching Opus-tier scores at Sonnet-tier prices

Anthropic released Claude Sonnet 5.5 on September 28, 2026, the second model in its Claude 5.5 family, and the headline number is not a benchmark score but a price tag: it costs exactly what Sonnet 5 cost, while Anthropic says it runs more than 30% faster and up to 30% cheaper per completed task.

What Changed

Sonnet 5.5 keeps the same $2-per-million-input-token, $10-per-million-output-token pricing as its predecessor, but the performance gap between the two is large. On Terminal-Bench 4.0, an agentic coding evaluation, Sonnet 5.5 scores 70.6% against Sonnet 5's 10.3% — a nearly sevenfold jump on a benchmark designed to test real coding agent behavior, not just code generation in isolation. On GDPval-AA, a test of real-world professional work across occupations, Sonnet 5.5 scores within two points of Anthropic's flagship Opus 5.5 model, despite being positioned as the cheaper, faster option.

Anthropic is also promoting Sonnet 5.5 as the first Sonnet-tier model with cybersecurity safeguards comparable to Opus 5.5, alongside improvements to long-horizon task handling, image understanding, and token efficiency that the company says reduces the number of tool calls needed to complete a given task.

Why This Matters

The positioning here is deliberate: Opus 5.5 for complex work requiring careful judgment, Sonnet 5.5 for well-scoped everyday tasks — bug fixes, documents, spreadsheets, and high-volume agentic workflows where cost per task compounds quickly. That split matters for any team running AI agents at scale, where the difference between a flagship model and a fast, cheap model often determines whether an AI-assisted workflow is economically viable at all.

Early enterprise deployments back up the efficiency claims. Epic Games reported that Sonnet 5.5 cleared the same quality bar expected from a higher-tier model across system audits and data reviews. Zendesk said the model resolved support tickets 20% faster than its predecessor. Base44 reported that across 118 real application builds, Sonnet 5.5 produced apps scoring level with Opus 5.

The Competitive Picture

The release lands squarely in an accelerating model-release cycle: Anthropic shipped Opus 5.5 shortly before Sonnet 5.5, and OpenAI has been iterating on its own GPT-6 family in the same window. For enterprise buyers evaluating cost per task rather than raw benchmark scores, Sonnet 5.5's pitch — Opus-adjacent performance on real-world work benchmarks at Sonnet-tier pricing — is designed to make the mid-tier model, not the flagship, the default choice for the bulk of production AI workloads.

Sonnet 5.5 is available now on the Claude Platform, Amazon Web Services, Google Cloud, and Microsoft Azure under the model ID claude-sonnet-5-5, with a zero-data-retention option for enterprise customers.