Claude Opus 5.5 Pricing: $4 / $20, 20% Below Opus 5
Anthropic's new Opus costs less than the model it replaces and, on the Arena leaderboard, scores above Claude Fable 5.1, which costs two and a half times as much.
Anthropic released Claude Opus 5.5 on 22 September 2026 at $4 per million input tokens and $20 per million output tokens. That is 20% below Claude Opus 5, which launched in July at $5 / $25. Cached input costs $0.20 per million, down from $0.50 on Opus 5, and batch requests are half price.
Claude Opus 5.5 at a glance
| API model ID | claude-opus-5-5 |
|---|---|
| Input | $4.00 per 1M tokens |
| Output | $20.00 per 1M tokens |
| Cached input | $0.20 per 1M tokens |
| Batch | 50% off |
| Fast mode | $8 / $40 per 1M tokens |
| Context window | 1M tokens |
| Maximum output | 128K tokens |
The live figures, with every other Opus model, are on our Claude Opus pricing page.
How it compares with the rest of the Claude line-up
| Model | Input / output per 1M | Cached input |
|---|---|---|
| Claude Fable 5.1 | $10.00 / $50.00 | $0.25 |
| Claude Opus 5.5 | $4.00 / $20.00 | $0.20 |
| Claude Opus 5 | $5.00 / $25.00 | $0.50 |
| Claude Sonnet 5 | $2.00 / $10.00 | $0.20 |
| Claude Haiku 4.5 | $1.00 / $5.00 | $0.10 |
Two things stand out. Opus 5.5 is now 60% cheaper than Claude Fable 5.1 on both input and output. And its cached-input price equals Sonnet 5's, which matters for agents: in a long coding session most input tokens are cache reads, so Opus costs less than twice as much as Sonnet there, even though its list prices are double.
On the Arena text leaderboard on 25 September, Opus 5.5 had a provisional score of 1509 from 2,307 votes, above Fable 5.1 at 1501. Provisional scores move as votes come in, and leaderboard positions are not the same as performance on your own tasks.
What it costs in practice
For a coding agent doing 400 tasks a month (about 750M input tokens with 85% served from cache, and 8M output tokens), Opus 5.5 costs $738 a month, against $1,081 on Opus 5 and $1,684 on Fable 5.1. On Sonnet 5, the same work costs $433.
For a chat workload without caching, such as 10M input and 2M output tokens a month, it costs $80.00, against $40.00 on GPT-6 Sol and $200 on GPT-6 Astra.
Against GPT-6 and Gemini
OpenAI launched GPT-6 Sol at $2 / $10 on the same day, half the price of Opus 5.5, and GPT-6 Astra, OpenAI's top model, costs $10 / $50. Google's Gemini 3.1 Pro costs $2 / $12 for prompts up to 200K tokens. Compare Opus 5.5 with GPT-6 Astra or with Gemini 3.1 Pro.
Should you switch?
- From Opus 5, 4.8 or earlier: yes, unless you depend on a specific older model's behaviour. Opus 5.5 is cheaper on every rate, and older Opus models stay available at $5 / $25 while you test.
- From Fable 5.1: test Opus 5.5 on your hardest tasks. If it holds up, the saving is 60%.
- From Sonnet 5: Opus costs twice as much. For cache-heavy agents, the real difference can be smaller; run your numbers in the agent cost calculator.
Sources: Anthropic pricing, TechCrunch.