Claude Sonnet 4.5 (latest)
Balanced Claude model for coding, analysis, agent workflows, and cost control
Shogo Agent Index
Supported#58 of 9238.7range 34.6–42.8 · 62% of weight measured
Agentic coding
38
57% measured
Long-horizon autonomy
54
80% measured
Knowledge-work agents
32
68% measured
Computer use
27
40% measured
Security & context learning
39
50% measured
Price
- Input
- $3.00/M
- Output
- $15.00/M
- Cache read
- $0.30/M
- 3:1 blend
- $6.00/M
- Agent-effective
- $1.39/M
Cheapest host for agents: Amazon Bedrock at $1.39/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 | 71.3% | 8% | source | |
| Terminal-Bench 2.0 | 46.5% | 8% | CAMEL-AI | |
| DeepSWE | est. | 6% | predicted 30 | |
| FrontierSWE | est. | 5% | predicted 6 | |
| CursorBench | est. | 4% | predicted 29 | |
| GSO-Bench | 14.7% | 4% | OpenHands | |
| METR Time Horizons | 67.4% | 10% | source | |
| Vending-Bench 2 | $3,839 | 6% | source | |
| PostTrainBench | est. | 4% | predicted 19 | |
| APEX-Agents | est. | 8% | predicted 29 | |
| GDPval | 42.5% | 7% | source | |
| Remote Labor Index | 2.1% | 5% | source | |
| DeepResearch Bench | 52.6% | 5% | source | |
| OSWorld 2.0 | est. | 6% | predicted 3 | |
| OSWorld | 62.9% | 4% | claude-sonnet-4-5-20250929 (100 steps) | |
| Cybench | 60.0% | 5% | source | |
| CL-bench | est. | 5% | predicted 18 |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| AlgoTune | Coding | 1.52× | source |
| ARC-AGI | Reasoning | 63.7% | source |
| ARC-AGI-2 | Reasoning | 13.6% | source |
| Chess Puzzles | Reasoning | 10.0% | source |
| CritPt | Reasoning | 1.1% | source |
| DTBench | Knowledge | 83.2% | source |
| FrontierMath Tiers 1-3 | Math | 23.9% | source |
| FrontierMath Tier 4 | Math | 2.4% | source |
| GeoBench | Multimodal | 60.0% | source |
| GPQA Diamond | Reasoning | 82.3% | source |
| Humanity's Last Exam | Reasoning | 13.7% | source |
| LMCA | Knowledge | 38.8% | source |
| MATH Level 5 | Math | 97.7% | source |
| OTIS Mock AIME | Math | 77.8% | source |
| ProofBench | Math | 19.0% | source |
| SciCode | Coding | 45.7% | source |
| SimpleBench | Reasoning | 54.3% | source |
| SimpleQA Verified | Knowledge | 30.7% | source |
| VPCT | Multimodal | 39.8% | source |
| WebDev Arena | Coding | 1391 | source |
API hosts
Live prices and uptime for every host serving Claude Sonnet 4.5 (latest) via OpenRouter.
| Host | Input | Output | Cache read | Agent $/M | Uptime 24h | Quant |
|---|---|---|---|---|---|---|
| Amazon Bedrock | $3.00 | $15.00 | $0.30 | $1.39 | 100.0% | — |
| Anthropic | $3.00 | $15.00 | $0.30 | $1.39 | 100.0% | — |
| Azure | $3.00 | $15.00 | $0.30 | $1.39 | — | — |
| Claude Platform on AWS | $3.00 | $15.00 | $0.30 | $1.39 | 100.0% | — |
| $3.00 | $15.00 | $0.30 | $1.39 | 100.0% | — |