anthropic
Anthropic: Claude Sonnet 5.5
anthropic/claude-sonnet-5.5
Our take
AI-assisted analysis based on the sources below. Suitability is an editorial assessment; test it on your own workload.
Anthropic's Claude Sonnet 5.5 (anthropic/claude-sonnet-5.5) is a Sonnet-class text-output model listed in the OpenRouter catalog as a direct successor to Claude Sonnet 5 for well-scoped everyday work. The catalog reports text, image, and file inputs, a one-million-token context window, mandatory reasoning with a configurable effort level, and support for tools and structured outputs. Design Arena entries recorded through OpenRouter cover agentic and generative tasks. No independent model card was available in this source pack, so claims rest on catalog data.[1]
Strengths
- According to OpenRouter's catalog description, Anthropic positions it as especially strong at building features and fixing bugs, framed as a direct upgrade over Claude Sonnet 5.[1]
- Catalog-reported capabilities include a 1,000,000-token context window and multimodal input of text, images, and files with text output.[1]
- API features listed by OpenRouter include mandatory reasoning with selectable effort levels, tool calling, and structured outputs.[1]
Limitations
- All evidence comes from the OpenRouter catalog; no independent model card was included in this source pack, so capabilities and benchmark entries should be validated on real workloads.[1]
- The catalog marks reasoning as mandatory, which could increase latency and token usage compared with non-reasoning modes; this implication is inferred, not measured.[1]
- Knowledge cutoff, license terms, and hardware requirements are not reported in the evidence.[1]
Where to use it
- Editorial suggestion: feature development and bug fixing in existing codebases, matching the vendor-described strengths reported by OpenRouter.[1]
- Editorial suggestion: agentic or design-oriented coding tasks such as code-to-video or UI component generation, where catalog-listed Design Arena entries exist; validate on your workload.[1]
- Editorial suggestion: document- or image-aware workflows that need very long context, given the cataloged input modalities and context window.[1]
Sources
- [1] OpenRouter catalog · Anthropic: Claude Sonnet 5.5Sources checked: Nov 13, 2026
About this model
Description below is the original text supplied by OpenRouter, shown in the source language.
Claude Sonnet 5.5 is Anthropic's Sonnet-class model for well-scoped everyday work, succeeding Claude Sonnet 5 as a direct upgrade. It is especially strong at building features, fixing bugs, and producing...
Specifications
- Model ID
- anthropic/claude-sonnet-5.5
- Canonical ID
- anthropic/claude-sonnet-5.5-20260928
- Creator
- anthropic
- Added to OpenRouter
- 28.09.2026
- Context length
- 1,000,000
- Modality
- text+image+file->text
- Input modalities
- text, image, file
- Output modalities
- text
- Token pricing
- USD per 1M tokens: in $2 / out $10
- Knowledge cutoff
- —
- Expiration date
- —
Tasks and capabilities
A grade appears only where the source supplies a matching measurement for that task, the model is active and the cohort holds at least 10 comparable models. The capability column lists facts the source declares — supported formats, reasoning and tool parameters — which are not measured quality.
| Task | Grade | Source-declared capability |
|---|---|---|
| General quality | SS — top 10% of the cohort | reasoning: mandatory · default high · efforts: max, xhigh, high, medium, low · parameters: include_reasoning, reasoning, reasoning_effort |
| Coding | No measurement | — |
| Agents | No measurement | parameters: structured_outputs, tool_choice, tools |
| Writing | No measurement | output: text |
| Reasoning | No measurement | reasoning: mandatory · default high · efforts: max, xhigh, high, medium, low · parameters: include_reasoning, reasoning, reasoning_effort |
| Document analysis | No measurement | input: file |
| Vision | No measurement | input: image |
| Image generation | No measurement | — |
| Audio | No measurement | — |
| Video | No measurement | — |
| Embeddings | No measurement | — |
Evaluations
| Source | Benchmark | Value | Grade | Data retrieved |
|---|---|---|---|---|
| Artificial Analysis | Intelligence index | 54.5 | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | agents · androidnativewin rate 58.6% · rank 6 | 1,260Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | agents · blender3dwin rate 59.9% · rank 3 | 1,268Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | agents · codetovideowin rate 70.9% · rank 2 | 1,318Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | agents · mobileappswin rate 57.6% · rank 5 | 1,261Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | agents · pdfwin rate 59.1% · rank 9 | 1,238Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | agents · spreadsheetwin rate 53.6% · rank 7 | 1,287Elo | AA — next 20% of the cohort | 13.11.2026 |
| Design Arena | agents · webappswin rate 58.5% · rank 4 | 1,284Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | models · 3dwin rate 54.7% · rank 7 | 1,402Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | models · codecategorieswin rate 56% · rank 8 | 1,347Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | models · datavizwin rate 64.3% · rank 3 | 1,360Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | models · gamedevwin rate 59.9% · rank 6 | 1,391Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | models · svgwin rate 56.7% · rank 6 | 1,303Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | models · uicomponentwin rate 64.8% · rank 2 | 1,387Elo | SS — top 10% of the cohort | 13.11.2026 |
| Design Arena | models · websitewin rate 53.6% · rank 10 | 1,311Elo | SS — top 10% of the cohort | 13.11.2026 |
How grades work
Each letter grade compares a model only against other models scored by the same source and metric in the current snapshot. S means the top 10%, A the next 20%, B the next 30%, C the rest. Models without a published measurement are shown as not rated — a supported format alone never implies measured quality.
- S — S — top 10% of the cohort
- A — A — next 20% of the cohort
- B — B — next 30% of the cohort
- C — C — below the B band
Providers
Optional live provider statistics from OpenRouter. Missing values are unknown, not zero.
| Provider | Context length | Price in /1M | Price out /1M | Latency p50 (30m, s) | Throughput p50 (30m, tok/s) | Uptime (30m) |
|---|---|---|---|---|---|---|
| 1,000,000 | $2 | $10 | — | — | 100% | |
| Amazon Bedrock | 1,000,000 | $2 | $10 | — | — | 100% |
| Azure | 1,000,000 | $2 | $10 | — | — | 90.48% |
| Claude Platform on AWS | 1,000,000 | $2 | $10 | — | — | 96.67% |
| Anthropic | 1,000,000 | $2 | $10 | — | — | 90% |
| Amazon Bedrock | 1,000,000 | $2.2 | $11 | — | — | 100% |
| Amazon Bedrock | 1,000,000 | $2.2 | $11 | — | — | 100% |
| 1,000,000 | $2.2 | $11 | — | — | 100% | |
| Azure | 1,000,000 | $2.2 | $11 | — | — | — |
| 1,000,000 | $2.2 | $11 | — | — | 100% |
Provider data retrieved: 13.11.2026
Related articles
- AWS shows how to keep AI agent searches within each user's permissions on Bedrock Managed Knowledge BaseA walkthrough builds a shared internal assistant on Amazon Bedrock AgentCore where a request interceptor, Cedar policies, ACL awareness and a resource-based policy keep each user's searches within the documents they may view.
- LLM batch APIs cut prices in half, but only for traffic that can waitA cost-modeling blog explains how batch discounts apply only to deferrable calls, why batch and caching stack multiplicatively, the billing bug this caused, and why dropping a model tier usually saves more.
- Google DeepMind and Artificial Analysis Introduce MKA, a Benchmark for Visually Grounded Agent WorkMKA is a 200-task benchmark that measures whether AI agents can extract decisive evidence from blueprints, scans, and charts and turn it into professional deliverables.
- Gemini 4 Carbon opens to general availability at launch pricingSimon Willison reports that Google's Gemini 4 Carbon, previously limited to a subset of cyber defenders, is now available to everyone, with launch prices set to double on December 1st.