Claude Sonnet 5.5
Claude Sonnet 5.5 by Anthropic: mid-tier workhorse model with fewer tool calls, near-flagship performance, and effort-based reasoning for coding and agents.
Model Overview
Capabilities, design details, and architectural traits
Claude Sonnet 5.5 - Anthropic's efficient mid-tier workhorse
Claude Sonnet 5.5 is Anthropic's faster, more efficient update to its mid-tier workhorse model. Its defining idea is completing a task with fewer tokens and fewer tool calls. In several of Anthropic's own tests it comes strikingly close to the flagship, Opus 5.5.
What sets it apart
| Trait | Detail |
|---|---|
| Leaner task completion | Finishes jobs with fewer tokens and fewer tool calls |
| Near-flagship performance | Comes strikingly close to Opus 5.5 in several of Anthropic's tests, narrowing the gap between the two tiers |
| Positioned for defined work | Best suited to well-defined everyday work: software debugging, coding, documents, presentations, spreadsheets, and interface design, while Opus 5.5 handles ambiguous or open-ended work |
| Effort-based reasoning | Runs at Low, Medium, or High effort; Claude Code and consumer apps default to Medium, the Claude Platform defaults to High |
| Leaner agent steps | Early customers report roughly one-third fewer tool calls and about half as many shell executions to finish coding jobs |
Efficiency as the product
Anthropic's pitch with Sonnet 5.5 is that raw step counts tell only part of the story. Every unnecessary tool call, failed action, and retry adds latency to an autonomous workflow. Sonnet 5.5 is built to complete the same jobs in fewer steps, which matters most for coding agents and long-running automated work.
Effort settings shape the tradeoff
Lower effort settings trade some reasoning depth for lower latency and token consumption. Higher settings let the model spend longer checking and refining its work. Anthropic reports that Sonnet 5.5 at Low or Medium effort can surpass Sonnet 5's best result, making effort selection a practical lever for production deployments.
Benchmark Performance
Independent evaluations · Artificial Analysis
Accuracy & Capability Details
Compare Models Side-by-Side
Evaluate specifications, pricing, and independent benchmark indices
| Model Details | |||
|---|---|---|---|
| General Info | |||
| Provider | Anthropic | Anthropic | Anthropic |
| Release Date | September 28, 2026 | September 22, 2026 | September 1, 2026 |
| Knowledge Cutoff | Jun 2026 | - | - |
| Context & Limits | |||
| Context Window | 1M | 1M | 1M |
| Pricing (per 1M tokens) | |||
| Input Pricing | $2 Best Input Pricing | $4 | $10 |
| Output Pricing | $10 Best Output Pricing | $20 | $50 |
| Modalities | |||
| Inputs | textimagefile | textimagefile | textimagefile |
| Outputs | text | text | text |
| Benchmarks (0-100) | |||
| Intelligence Index | 56.0 | 57.6 Best Intelligence Index | 53.4 |
| Coding Index | - | - | 81.6 |
| Agentic Index | - | - | 57.9 |
Humanity's Last Exam
Extremely difficult logical reasoning and knowledge.
Long Context Reasoning
Logical reasoning over long context windows.
SciCode Benchmark
Scientific coding and mathematical modeling.
Independent evaluation data provided by Artificial Analysis. To view the latest benchmarks and full details, visit their official site.
Explore more from Anthropic
Other models by Anthropic
Claude Opus 5.5
Claude Fable 5.1
Claude Opus 5
Claude Fable 5
Top AI Models
Leading alternatives by intelligence score
