Agentic coding performance
Anthropic reports a 70.6% score on Terminal-Bench 4.0, compared with 10.3% for Sonnet 5. The evaluation measures multi-step professional tasks completed in a command-line interface.
Claude Sonnet 5.5
Claude Sonnet 5.5 is Anthropic's AI model for well-scoped coding, knowledge work, document creation, spreadsheets, and computer-interface tasks. It is designed for users who need faster iteration and lower per-task costs than Claude Sonnet 5, while more complex open-ended work may be better suited to Claude Opus 5.5.
Claude Sonnet 5.5 overview
Claude Sonnet 5.5 is Anthropic’s second model in the Claude 5.5 family and a faster, lower-cost complement to Claude Opus 5.5. It is designed for well-scoped everyday work, bug fixing, and creating polished documents, presentations, and spreadsheets, with additional strength in design-oriented tasks and image understanding.
Compared with Claude Sonnet 5, Anthropic reports more than 30% faster output generation, substantially higher scores on several coding evaluations, and up to 30% lower cost per task in its testing. Opus 5.5 remains the better fit for complex, open-ended work that requires sustained judgment, while Sonnet 5.5 is positioned for faster iteration on less complex tasks.
Key features and evaluated capabilities
Code changes across complex repositories
On FrontierCode at High effort, Sonnet 5.5 scores 46.2%; on CursorBench 4.0, it scores 55.5%. These evaluations cover merge-worthy code changes and ambiguous, multi-file tasks from real Cursor sessions.
Knowledge-work support
The model scores 1844 on GDPval-AA v2.1 and 1811 on AA-Briefcase v1.1 in the published comparison. Anthropic describes these evaluations as covering real-world and long-horizon knowledge work.
Computer use and image understanding
Sonnet 5.5 scores 80.1% partial on OSWorld 2.1 for computer use and 61.6% without tools on Chartography for visual chart recognition. Anthropic also highlights image understanding and screenshot-based task performance.
Faster iteration
Anthropic says Sonnet 5.5 generates outputs more than 30% faster than Sonnet 5. It also reports that the model often batches tool calls, which can reduce the number of steps needed for coding tasks.
Lower cost per completed task
The model has the same listed token rates as Sonnet 5, while Anthropic reports up to 30% lower cost per task in its testing because Sonnet 5.5 typically uses fewer tokens for the same work.
Practical use cases
Software development and bug fixing
Use Sonnet 5.5 to inspect repositories, make multi-file changes, fix bugs, and work through command-line tasks. Its reported coding results and faster iteration are relevant when engineers need frequent feedback during implementation.
Structured knowledge work
Use the model for well-defined research, analysis, and document-production tasks where the output can be reviewed and refined. Anthropic’s knowledge-work evaluations include both occupational tasks and longer-horizon work.
Business content production
Generate or revise polished documents, presentations, and spreadsheets, then iterate quickly on wording, structure, or formatting.
Visual and computer-interface tasks
Apply the model to workflows that involve screenshots, charts, or computer interfaces. Anthropic reports results for computer use, visual chart recognition, and screenshot-based task performance.
Pros and Cons
Pros
- More than 30% faster output generation than Sonnet 5, according to Anthropic.
- Strong reported gains on agentic coding and long-horizon knowledge-work evaluations.
- Supports work involving documents, slides, spreadsheets, computer use, and image or chart understanding.
- Anthropic reports up to 30% lower cost per task than Sonnet 5 in its testing.
- Launches with cybersecurity safeguards and fallbacks associated with more capable models.
Cons
- Benchmark results do not establish performance for every workflow; Anthropic notes that scores capture only one facet of model capability.
- The source does not provide detailed setup, workflow, team-administration, or deployment guidance specific to Sonnet 5.5.
- Cybersecurity and biology safeguards target a narrow set of high-risk requests, so users remain responsible for evaluating outputs and managing access in their own workflows.
FAQ
What is Claude Sonnet 5.5 best suited for?
Claude Sonnet 5.5 is intended for well-scoped everyday tasks, bug fixing, and producing polished documents, slides, and spreadsheets. Anthropic positions Claude Opus 5.5 as the stronger option for complex, open-ended work requiring sustained judgment.
How much faster is Sonnet 5.5 than Sonnet 5?
Anthropic reports that Sonnet 5.5 generates outputs more than 30% faster than Sonnet 5. It is described as the fastest Sonnet model to date.
How is Claude Sonnet 5.5 priced?
Anthropic lists API rates of $2 per million input tokens, $10 per million output tokens, and $0.20 per million tokens for cache reads. The company says the model typically uses fewer tokens and costs up to 30% less per task than Sonnet 5 in its testing.
Where can Claude Sonnet 5.5 be used?
The model is available through Anthropic’s Claude offerings and API pricing. The Claude pricing page lists web, desktop, and mobile access across plans, while specific model availability can vary by plan and product.
What safety measures does Sonnet 5.5 include?
Anthropic says Sonnet 5.5 launched with cybersecurity safeguards and fallbacks because its cybersecurity capabilities are comparable to Opus 5’s. Its biology safeguards are the same as Sonnet 5’s, and the safeguards target a narrow set of high-risk requests.
Quick Facts
- Category
- AI model / developer tool
- Provider
- Anthropic
- Model family
- Claude 5.5
- Primary tasks
- Coding, knowledge work, documents, slides, spreadsheets, and computer use
- Reported output speed
- 30%+ faster than Sonnet 5
- API rates
- $2 per million input tokens; $10 per million output tokens; $0.20 per million cache-read tokens
Alternatives à Claude Sonnet 5.5
Grok 4.7
Grok 4.7 is an xAI model for software development and knowledge work, including extended coding tasks and professional analysis. It is available through Grok Build, Cursor, the Grok API, and other developer platforms, with token pricing starting at $2 per million input tokens and $6 per million output tokens.
ByteAsk
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
MeetStream
MeetStream is a meeting bot API for Zoom, Google Meet, Microsoft Teams, and Webex. It helps developers record, stream, and analyze meetings programmatically, with usage-based pricing and a $5 free credit to start.
CreateOS Sandbox
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
GitBot
GitBot is a local browser-based workspace for creating reusable coding bots powered by Claude Code, Codex, or OpenCode. It helps developers run those agents across repository folders with persistent threads and configurable permissions.
hob
hob is an independent workspace for coding agents that keeps agent sessions, terminals, history, and follow-up work organized around the tools and providers you already use. It is aimed at developers who want local control over routing, history, and workspace structure rather than a bundled model stack.