Claude Opus 4.8
Top Claude Opus tier for the hardest reasoning, coding, and long-horizon agents
Shogo Agent Index
Supported#20 of 9256.3range 50.9–61.8 · 44% of weight measured
Agentic coding
63
29% measured
Long-horizon autonomy
66
50% measured
Knowledge-work agents
43
72% measured
Computer use
47
60% measured
Security & context learning
55
estimated
Price
- Input
- $5.00/M
- Output
- $25.00/M
- Cache read
- $0.50/M
- 3:1 blend
- $10.00/M
- Agent-effective
- $2.31/M
Cheapest host for agents: Amazon Bedrock at $2.31/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 83 | |
| Terminal-Bench 2.0 | est. | 8% | predicted 86 | |
| DeepSWE | 59.0% | 6% | mini-swe-agent · $13.22/task | |
| FrontierSWE | est. | 5% | predicted 29 | |
| CursorBench | est. | 4% | predicted 44 | |
| GSO-Bench | 47.1% | 4% | OpenHands | |
| METR Time Horizons | est. | 10% | predicted 80 | |
| Vending-Bench 2 | $5,787 | 6% | source | |
| PostTrainBench | 31.0% | 4% | Claude Code | |
| APEX-Agents | 48.9% | 8% | source | |
| GDPval | est. | 7% | predicted 57 | |
| Remote Labor Index | 8.3% | 5% | source | |
| DeepResearch Bench | 50.2% | 5% | source | |
| OSWorld 2.0 | 20.6% | 6% | source | |
| OSWorld | est. | 4% | predicted 87 | |
| Cybench | est. | 5% | predicted 85 | |
| CL-bench | est. | 5% | predicted 25 |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| ARC-AGI | Reasoning | 92.5% | source |
| ARC-AGI-2 | Reasoning | 72.1% | source |
| Chess Puzzles | Reasoning | 52.0% | source |
| CritPt | Reasoning | 20.9% | source |
| DTBench | Knowledge | 94.9% | source |
| FrontierCode | Coding | 46.5% | claude-code |
| FrontierMath Tiers 1-3 | Math | 80.0% | source |
| FrontierMath Tier 4 | Math | 56.1% | source |
| GeoBench | Multimodal | 81.0% | source |
| GPQA Diamond | Reasoning | 91.0% | source |
| LMCA | Knowledge | 57.5% | source |
| MirrorCode | Coding | 65.6% | source |
| OTIS Mock AIME | Math | 98.3% | source |
| ProofBench | Math | 69.0% | source |
| SciCode | Coding | 53.5% | source |
| SimpleBench | Reasoning | 64.8% | source |
| SimpleQA Verified | Knowledge | 53.0% | source |
| WebDev Arena | Coding | 1559 | source |
API hosts
Live prices and uptime for every host serving Claude Opus 4.8 via OpenRouter.
| Host | Input | Output | Cache read | Agent $/M | Uptime 24h | Quant |
|---|---|---|---|---|---|---|
| Amazon Bedrock | $5.00 | $25.00 | $0.50 | $2.31 | 99.0% | — |
| Anthropic | $5.00 | $25.00 | $0.50 | $2.31 | 100.0% | — |
| Azure | $5.00 | $25.00 | $0.50 | $2.31 | 100.0% | — |
| Claude Platform on AWS | $5.00 | $25.00 | $0.50 | $2.31 | 100.0% | — |
| $5.00 | $25.00 | $0.50 | $2.31 | 100.0% | — | |
| Anthropic | $10.00 | $50.00 | $1.00 | $4.62 | 100.0% | — |