Claude Opus 5.5
Claude model for long-running agentic coding and knowledge work
Shogo Agent Index
Supported#3 of 9270.9range 64.8–76.9 · 32% of weight measured
Agentic coding
81
26% measured
Long-horizon autonomy
76
50% measured
Knowledge-work agents
58
52% measured
Computer use
66
estimated
Security & context learning
62
estimated
Price
- Input
- $4.00/M
- Output
- $20.00/M
- Cache read
- $0.20/M
- 3:1 blend
- $8.00/M
- Agent-effective
- $1.69/M
Cheapest host for agents: Amazon Bedrock at $1.69/M.
Agent-effective = 20 input tokens per output token, 85% cache hits. Why.
Agentic benchmarks
| Benchmark | Score | Normalised | Index weight | Harness / source |
|---|---|---|---|---|
| SWE-bench Verified | est. | 8% | predicted 89 | |
| Terminal-Bench 2.0 | est. | 8% | predicted 96 | |
| DeepSWE | est. | 6% | predicted 82 | |
| FrontierSWE | 62.3% | 5% | proximus · $98.87/task | |
| CursorBench | 57.8% | 4% | source · $13.43/task | |
| GSO-Bench | est. | 4% | predicted 78 | |
| METR Time Horizons | est. | 10% | predicted 87 | |
| Vending-Bench 2 | $9,235 | 6% | source | |
| PostTrainBench | 43.8% | 4% | Claude Code | |
| APEX-Agents | 73.5% | 8% | source | |
| GDPval | est. | 7% | predicted 72 | |
| Remote Labor Index | 21.3% | 5% | source | |
| DeepResearch Bench | est. | 5% | predicted 53 | |
| OSWorld 2.0 | est. | 6% | predicted 46 | |
| OSWorld | est. | 4% | predicted 96 | |
| Cybench | est. | 5% | predicted 94 | |
| CL-bench | est. | 5% | predicted 31 |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| ARC-AGI | Reasoning | 98.5% | source |
| ARC-AGI-2 | Reasoning | 93.3% | source |
| Chess Puzzles | Reasoning | 67.0% | source |
| CritPt | Reasoning | 31.7% | source |
| DTBench | Knowledge | 98.9% | source |
| FrontierCode | Coding | 54.6% | claude-code |
| FrontierMath Tiers 1-3 | Math | 91.2% | source |
| FrontierMath Tier 4 | Math | 95.0% | source |
| GPQA Diamond | Reasoning | 94.1% | source |
| LMCA | Knowledge | 68.2% | source |
| MirrorCode | Coding | 95.6% | source |
| OTIS Mock AIME | Math | 100.0% | source |
| ProofBench | Math | 100.0% | source |
| SciCode | Coding | 66.9% | source |
| SimpleBench | Reasoning | 88.4% | source |
| SimpleQA Verified | Knowledge | 72.2% | source |
| WebDev Arena | Coding | 1820 | source |
| WeirdML v3 | Coding | 31.2% | claude_code |
API hosts
Live prices and uptime for every host serving Claude Opus 5.5 via OpenRouter.
| Host | Input | Output | Cache read | Agent $/M | Uptime 24h | Quant |
|---|---|---|---|---|---|---|
| Amazon Bedrock | $4.00 | $20.00 | $0.20 | $1.69 | 100.0% | — |
| Anthropic | $4.00 | $20.00 | $0.20 | $1.69 | 100.0% | — |
| Azure | $4.00 | $20.00 | $0.20 | $1.69 | 100.0% | — |
| Claude Platform on AWS | $4.00 | $20.00 | $0.20 | $1.69 | 100.0% | — |
| $4.00 | $20.00 | $0.20 | $1.69 | 100.0% | — | |
| Anthropic | $8.00 | $40.00 | $0.40 | $3.37 | 100.0% | — |