AnthropicAPI Host Benchmarks
Claude Sonnet 5 — API Providers & Speed Matrix
Compare live throughput, TTFT latency distributions, prompt cache pricing, and hardware stacks across cloud hosting endpoints.
Fastest Provider
1200 t/s
Lowest Latency (TTFT)
0.09s P50
Best Price Value
$0.9 / $5
Blended Pricing Workload Calculator
Switch between standard conversational ratios, agentic caching loops, and heavy RAG workloads.
| Provider & Host | Hardware Stack | Output Speed (TPS) | Latency TTFT (P50) | Input Price ($/1M) | Output Price ($/1M) | Cache Read ($/1M) | Blended Price ($/1M) | SLA |
|---|---|---|---|---|---|---|---|---|
AnthropicOfficial | Cloud TPU / Nvidia H100 | 96 t/s | 197.81s (276.93s p90) | $2 | $10 | $0.2 | $4 | 99.99% |
Groq Cloud | Groq LPU v2 (SRAM) | 403 t/s | 0.09s (0.12s p90) | $1.7 | $8.5 | $0.16 | $3.4 | 99.95% |
Cerebras Cloud | CS-3 Wafer-Scale Engine | 1200 t/s | 0.12s (0.16s p90) | $1.8 | $9 | — | $3.6 | 99.90% |
Together AI | Nvidia H100 SXM5 | 144 t/s | 158.25s (217.59s p90) | $1.3 | $7 | $0.12 | $2.73 | 99.95% |
DeepInfra | Nvidia H100 / L40S | 125 t/s | 178.03s (247.26s p90) | $0.9 | $5 | $0.1 | $1.93 | 99.90% |
Fireworks AI | Nvidia B200 / H100 | 182 t/s | 138.47s (187.92s p90) | $1.2 | $6.5 | $0.12 | $2.52 | 99.99% |