MicrosoftAPI Host BenchmarksEstimated
Phi 4 Mini — API Providers & Speed Matrix
Modelled throughput, TTFT latency and prompt-cache pricing estimates across cloud hosting endpoints, derived from official list prices and public benchmarks.
Fastest provider
750 t/s
Lowest latency (TTFT)
90 ms P50
Best price
$0.07 / $0.3
Blended Pricing Workload Calculator
Switch between standard conversational ratios, agentic caching loops, and heavy RAG workloads.
| Provider & Host | Hardware Stack | Output Speed (TPS) | Latency TTFT (P50) | Input Price ($/1M) | Output Price ($/1M) | Cache Read ($/1M) | Blended Price ($/1M) | SLA |
|---|---|---|---|---|---|---|---|---|
MicrosoftOfficial | Cloud TPU / Nvidia H100 | 60 t/s | 800 ms (1.12 s p90) | $0.15 | $0.6 | $0.015 | $0.26 | 99.99% |
| Groq LPU v2 (SRAM) | 252 t/s | 90 ms (120 ms p90) | $0.13 | $0.51 | $0.012 | $0.23 | 99.95% | |
| CS-3 Wafer-Scale Engine | 750 t/s | 120 ms (160 ms p90) | $0.14 | $0.54 | — | $0.24 | 99.90% | |
| Nvidia H100 SXM5 | 90 t/s | 640 ms (880 ms p90) | $0.1 | $0.42 | $0.009 | $0.18 | 99.95% | |
| Nvidia H100 / L40S | 78 t/s | 720 ms (1.00 s p90) | $0.07 | $0.3 | $0.007 | $0.13 | 99.90% | |
| Nvidia B200 / H100 | 114 t/s | 560 ms (760 ms p90) | $0.09 | $0.39 | $0.009 | $0.17 | 99.99% |
Figures are modelled estimates based on published list prices and public throughput benchmarks — not measured on live endpoints. Verify current pricing and terms with each host before committing.