Text and chat · Anthropic
Claude
Anthropic's models, strong at writing, code and long agent work, with vision and 1M context, from the small Haiku to the top Fable. Anonymized, and mostly on the dear side.
Anonymized High costWhat sets it apart
Claude pairs a 1M token window and 128K output with adaptive thinking that always runs on the top tiers, and every current model reads images. Four tiers cover the range: Haiku 5.5 at $0.10 per million input tokens, Sonnet 5.5 at $2, Opus 5.5 at $4, and Fable 5.1 at $10.
- 1M context, 128K output
- Adaptive thinking, effort setting
- Four tiers: Haiku, Sonnet, Opus, Fable
Recommended for: write, summarize and answer questions every day ; write or fix code.
Strengths and limits
Where it shines
- Clear writing and polished documents (Sonnet 5.5)
- Strong at coding and long multi-step agent tasks
- Reads images, several per message, with tool calling
- Haiku 5.5 is the cheap, fast tier for classification and extraction
Think twice when
- Anonymized, not private: Anthropic's retention policy applies
- Dear per token apart from Haiku; Fable and Fast versions are among the dearest
- 1M token context (4.5 versions: 198K)
- Vision, tool calling, adjustable reasoning effort
Use cases
Polishing a document
Rewrite this project update for a non-technical board: keep it under 200 words, lead with the decision we need, and keep every number.
Sonnet 5.5 is Anthropic’s speed and quality pick, at $2 per million input tokens.
A long coding task
Read the repository I attached, add pagination to the orders list, update the tests, and tell me what you changed.
Opus 5.5 is built for long agentic coding and knowledge work.
Questions about a scanned page
This is a photo of a restaurant receipt. Split the total among four people, with the wine only on two of them.
Every current Claude model accepts images.
Against its rivals
- GPTGPT-6 Astra scores 53 on the index against 58 for Opus 5.5, with a different price ladder.#4 of 90 on Artificial Analysis Intelligence Index (53 points)
- GeminiGemini 4 Argon reaches 53 on the index and Gemini Flash tiers cost less.#5 of 90 on Artificial Analysis Intelligence Index (53 points)
- GrokGrok 4.7 scores 46 on the index, well below Opus 5.5 at 58.#12 of 90 on Artificial Analysis Intelligence Index (46 points)
Version history
Each version, dated, with what it changed over the one before.
Hollow points are earlier versions no longer offered in Sub Rosa.
See the numbers
| Version | Released | points |
|---|---|---|
| Claude Opus 4.5 | 2025-11-24 | 24 |
| Claude Opus 4.6 | 2026-02-05 | 26 |
| Claude Sonnet 4.6 | 2026-02-17 | 25 |
| Claude Opus 4.7 | 2026-04-16 | 41 |
| Claude Opus 4.8 | 2026-05-28 | 42 |
| Claude Fable 5 | 2026-06-09 | 50 |
| Claude Sonnet 5 | 2026-06-30 | 38 |
| Claude Opus 5 | 2026-07 | 51 |
| Claude Fable 5.1 | 2026-09-01 | 53 |
| Claude Opus 5.5 | 2026-09-22 | 58 |
| Claude Sonnet 5.5 | 2026-09-28 | 56 |
| Claude Haiku 5.5 | 2026-10-07 | 43 |
- Claude Sonnet 4.5
- Anthropic’s coding and agent model of the time: 77.2% on SWE-bench Verified, 200K context, $3 / $15 per million tokens.
- Claude Opus 4.5
- Opus tier at $5 / $25 per million tokens, 200K context.
- Claude Opus 4.6
- Context window grows from 200K to 1M tokens.
- Claude Sonnet 4.6
- Sonnet gets the 1M token context window.
- Claude Opus 4.7
- Introduces the current tokenizer, per Anthropic’s documentation.
- Claude Opus 4.8
- Scores 42 on the Artificial Analysis index, up from 41 for Opus 4.7.
- Claude Fable 5
- New top tier above Opus, priced at $10 / $50 per million tokens.
- Claude Sonnet 5
- Fifth-generation Sonnet at $2 / $10 per million tokens, with reasoning.
- Claude Opus 5
- Opus at $5 / $25 per million tokens, scoring 51 on the Artificial Analysis index.
- Claude Fable 5.1
- Refresh of the top tier, scoring 53 on the index against 50 for Fable 5.
- Claude Opus 5.5
- About 40% lower cost on typical workloads and 30% faster output than Opus 5, with a list price cut to $4 / $20.
- Better agentic coding and knowledge work: Terminal-Bench 4.0 at 66.4% against 52.3% for Opus 5.
- Claude Sonnet 5.5
- Scores 56 on the Artificial Analysis index, up from 38 for Sonnet 5, at the same $2 / $10 price.
- Adaptive thinking with a lowest setting that turns off up-front thinking.
- Claude Haiku 5.5Latest
- Smallest and fastest current tier, with the same 1M context, 128K output and adaptive thinking as the larger models. Scores 43 on the Artificial Analysis index at its highest effort, against 56 for Sonnet 5.5.
- Anthropic aims it at high-volume, latency-sensitive work such as classification, extraction, routing and subagent tasks.
Benchmarks
How to read these scores: what a benchmark tells you.
Artificial Analysis Intelligence Index: #1 of 90 models measured in the catalog. One score averaging about ten independent tests of reasoning, knowledge, math, coding and agent work, all run the same way for every model. Higher is better. Each model is listed at its highest reasoning effort. Two or three points apart is a close call; ten points is a different class of model.
See the numbers
| Model | points |
|---|---|
| 1. Claude Opus 5.5 | 58 |
| 2. Claude Sonnet 5.5 | 56 |
| 3. Claude Fable 5.1 | 53 |
| 4. GPT-6 Astra | 53 |
| 5. Gemini 4 Argon | 53 |
| 6. GPT-6.1 Sol | 52 |
| 7. Muse Wonder | 52 |
| 10. GPT-6 Sol | 48 |
Announced by the maker
Measured by the maker with its own settings: useful, but not directly comparable with the independent scores above.
| Benchmark | Model | Score | Source |
|---|---|---|---|
| swe-bench-verified | Claude Sonnet 4.5 | 77.2 | anthropic.com |
All 16 models in this family
| Model | Use | Privacy | Price |
|---|---|---|---|
Claude Sonnet 5.5claude-sonnet-5-5 | Text | Anonymized | $4.496 per million tokens written$0.00315 a page written |
Claude Sonnet 5claude-sonnet-5 | Text | Anonymized | $4.496 per million tokens written$0.00315 a page written |
Claude Sonnet 4.6claude-sonnet-4-6 | Text | Anonymized | $6.744 per million tokens written$0.00472 a page written |
Claude Sonnet 4.5claude-sonnet-4-5 | Text | Anonymized | $6.744 per million tokens written$0.00472 a page written |
Claude Opus 5.5claude-opus-5-5 | Text | Anonymized | $8.992 per million tokens written$0.00629 a page written |
Claude Opus 5.5 Fastclaude-opus-5-5-fast | Text | Anonymized | $17.98 per million tokens written$0.0126 a page written |
Claude Opus 5claude-opus-5 | Text | Anonymized | $11.24 per million tokens written$0.00787 a page written |
Claude Opus 5 Fastclaude-opus-5-fast | Text | Anonymized | $22.48 per million tokens written$0.0157 a page written |
Claude Opus 4.8claude-opus-4-8 | Text | Anonymized | $11.24 per million tokens written$0.00787 a page written |
Claude Opus 4.8 Fastclaude-opus-4-8-fast | Text | Anonymized | $22.48 per million tokens written$0.0157 a page written |
Claude Opus 4.7claude-opus-4-7 | Text | Anonymized | $11.24 per million tokens written$0.00787 a page written |
Claude Opus 4.6claude-opus-4-6 | Text | Anonymized | $11.24 per million tokens written$0.00787 a page written |
Claude Opus 4.5claude-opus-4-5 | Text | Anonymized | $11.24 per million tokens written$0.00787 a page written |
Claude Fable 5claude-fable-5 | Text | Anonymized | $22.48 per million tokens written$0.0157 a page written |
Claude Fable 5.1claude-fable-5-1 | Text | Anonymized | $22.48 per million tokens written$0.0157 a page written |
Claude Haiku 5.5claude-haiku-5-5 | Text | Anonymized | $0.225 per million tokens written$0.000157 a page written |
Specifications
| Specification | Claude Sonnet 5.5 | Claude Opus 5.5 | Claude Fable 5.1 | Claude Haiku 5.5 |
|---|---|---|---|---|
| Released | September 28, 2026 | September 22, 2026 | September 1, 2026 | October 7, 2026 |
| Weights | Closed | Closed | Closed | · |
| Accepts | text, images | text, images | text, images | text, images |
| Produces | text | text | text | text |
| Context | 1,000K tokens | 1,000K tokens | 1,000K tokens | 1,000K tokens |
| Longest answer | 128K tokens | 128K tokens | 128K tokens | 128K tokens |
| Abilities | Reads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for code | Reads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for code | Reads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for code | Reads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for code |
| Price | $4.496 per million tokens written | $8.992 per million tokens written | $22.48 per million tokens written | $0.225 per million tokens written |
| Privacy | Anonymized | Anonymized | Anonymized | Anonymized |
Prices, context and lengths from the live catalog on October 16, 2026; parameters and licenses from the maker.
Sources
Everything on this page comes from these pages, read on October 16, 2026 or as dated on each score.
Show the 23 sources
- platform.claude.com/docs/en/models/opus-5-5/overview
- platform.claude.com/docs/en/models/sonnet-5-5/overview
- platform.claude.com/docs/en/models/overview
- anthropic.com/news/claude-opus-5-5
- anthropic.com/news/claude-sonnet-4-5
- artificialanalysis.ai/models/claude-sonnet-5-5
- artificialanalysis.ai/models/claude-opus-5-5
- artificialanalysis.ai/models/claude-fable-5-1
- artificialanalysis.ai/models/claude-fable-5
- artificialanalysis.ai/models/claude-opus-5
- artificialanalysis.ai/models/claude-sonnet-5
- artificialanalysis.ai/models/claude-opus-4-8
- artificialanalysis.ai/models/claude-opus-4-7
- artificialanalysis.ai/models/claude-opus-4-6
- artificialanalysis.ai/models/claude-opus-4-5
- artificialanalysis.ai/models/claude-sonnet-4-6
- artificialanalysis.ai/leaderboards/models
- aiweekly.co/fr/alerts/anthropic-sort-claude-opus-55-40-moins-cher-quopus-5
- tech.ifeng.com/c/8wp16Vi4lWz
- thezvi.substack.com/p/claude-fable-5-and-mythos-5-the-system
- platform.claude.com/docs/en/models/haiku-5-5/overview
- anthropic.com/claude-haiku-5-5
- artificialanalysis.ai/models/claude-haiku-5-5