Claude Opus 5.5
Anthropic's current default model: long-running agentic coding and knowledge work at $4 / $20 per million tokens.
What it is
Claude Opus 5.5 is the first model in Anthropic's Claude 5.5 family. Anthropic's own docs say to start with Opus 5.5 for most workloads and move up to Fable 5.1 only when evals at higher effort still fall short. It replaced Claude Opus 5 at a 20% lower list price and a 60% lower cache-read price.
Pricing
| Input | $4.00 |
| Output | $20.00 |
| Cached input (cache read) | $0.200 |
| Cache write | $5.00 |
| Batch input | $2.00 |
| Batch output | $10.00 |
| Fast / priority input | $8.00 |
| Fast / priority output | $40.00 |
- Cache read is 5% of input price (0.05x).
- 1-hour cache write is $8 / MTok.
- Fast mode (research preview, Claude API only) is $8 / $40.
- Full 1M context at standard price; no long-context surcharge.
Capability highlights
- Anthropic says it performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5 at default settings.
- Output generation is more than 30% faster than Opus 5 (vendor claim).
- Thinking can no longer be switched off; forced tool use returns an error.
- Attempts to circumvent containment boundaries are about 85% less frequent than Opus 5 (vendor claim).
Modalities: text, image → text. Adaptive thinking, always on; default effort medium. 300K output tokens on the Message Batches API with a beta header.
How to use it
API: Claude API model ID claude-opus-5-5; Amazon Bedrock anthropic.claude-opus-5-5; Google Cloud, Microsoft Foundry and Claude Platform on AWS use claude-opus-5-5.
Versus Claude Opus 5
Opus 5.5 is $4 / $20 vs Opus 5 at $5 / $25; cache reads $0.20 vs $0.50.
| Claude Opus 5 | Opus 5.5 | |
|---|---|---|
| Input / 1M | $5.00 | $4.00 |
| Output / 1M | $25.00 | $20.00 |
| Cached input / 1M | $0.500 | $0.200 |
| Context window | 1M | 1M |
| Max output | 128K | 128K |
Published benchmark numbers
Vendor-reported, quoted from the linked announcement. Not independently verified; different vendors run benchmarks under different settings.
| Benchmark | Opus 5.5 | Comparison in the same source | Source |
|---|---|---|---|
| Terminal-Bench 4.0 | 66.4% | Opus 5: 52.3%, Fable 5.1: 55.8% | Anthropic announcement |
| GDPval-AA v2.1 (Elo) | 1846 | Opus 5: 1708, Fable 5.1: 1735 | Anthropic announcement |
| OSWorld 2.1 (partial) | 81.8% | Opus 5: 74.0% | Anthropic announcement |
| Humanity's Last Exam (with tools) | 67.7% | Opus 5: 63.6% | Anthropic announcement |
Comparisons
Sources
Every number above comes from one of these pages, read on the date shown. Searched for: opus 5.5, opus5.5, claude opus 5.5.
- Claude Opus 5.5 model page (platform.claude.com) read 2026-10-08
release date, model IDs, context, output, pricing, cutoff - Anthropic pricing page read 2026-10-08
cache, batch, fast mode prices - Anthropic announcement: Claude Opus 5.5 read 2026-10-08
benchmarks (vendor-reported), availability, claims vs Opus 5