AlibabaAPI Host Benchmarks
Qwen3.7 Plus — API Providers & Speed Matrix
Compare live throughput, TTFT latency distributions, prompt cache pricing, and hardware stacks across cloud hosting endpoints.
Fastest Provider
675 t/s
Lowest Latency (TTFT)
0.09s P50
Best Price Value
$0.18 / $0.8
Blended Pricing Workload Calculator
Switch between standard conversational ratios, agentic caching loops, and heavy RAG workloads.
| Provider & Host | Hardware Stack | Output Speed (TPS) | Latency TTFT (P50) | Input Price ($/1M) | Output Price ($/1M) | Cache Read ($/1M) | Blended Price ($/1M) | SLA |
|---|---|---|---|---|---|---|---|---|
AlibabaOfficial | Cloud TPU / Nvidia H100 | 54 t/s | 2.35s (3.29s p90) | $0.4 | $1.6 | $0.04 | $0.7 | 99.99% |
Groq Cloud | Groq LPU v2 (SRAM) | 227 t/s | 0.09s (0.12s p90) | $0.34 | $1.36 | $0.032 | $0.59 | 99.95% |
Cerebras Cloud | CS-3 Wafer-Scale Engine | 675 t/s | 0.12s (0.16s p90) | $0.36 | $1.44 | — | $0.63 | 99.90% |
Together AI | Nvidia H100 SXM5 | 81 t/s | 1.88s (2.59s p90) | $0.26 | $1.12 | $0.024 | $0.48 | 99.95% |
DeepInfra | Nvidia H100 / L40S | 70 t/s | 2.12s (2.94s p90) | $0.18 | $0.8 | $0.02 | $0.34 | 99.90% |
Fireworks AI | Nvidia B200 / H100 | 103 t/s | 1.65s (2.23s p90) | $0.24 | $1.04 | $0.024 | $0.44 | 99.99% |