AlibabaAPI Host Benchmarks
Qwen3.8 27B (Non-reasoning) — API Providers & Speed Matrix
Compare live throughput, TTFT latency distributions, prompt cache pricing, and hardware stacks across cloud hosting endpoints.
Fastest Provider
775 t/s
Lowest Latency (TTFT)
0.09s P50
Best Price Value
$0.23 / $1.5
Blended Pricing Workload Calculator
Switch between standard conversational ratios, agentic caching loops, and heavy RAG workloads.
| Provider & Host | Hardware Stack | Output Speed (TPS) | Latency TTFT (P50) | Input Price ($/1M) | Output Price ($/1M) | Cache Read ($/1M) | Blended Price ($/1M) | SLA |
|---|---|---|---|---|---|---|---|---|
AlibabaOfficial | Cloud TPU / Nvidia H100 | 62 t/s | 3.06s (4.28s p90) | $0.5 | $3 | $0.05 | $1.13 | 99.99% |
Groq Cloud | Groq LPU v2 (SRAM) | 260 t/s | 0.09s (0.12s p90) | $0.42 | $2.55 | $0.04 | $0.95 | 99.95% |
Cerebras Cloud | CS-3 Wafer-Scale Engine | 775 t/s | 0.12s (0.16s p90) | $0.45 | $2.7 | — | $1.01 | 99.90% |
Together AI | Nvidia H100 SXM5 | 93 t/s | 2.45s (3.37s p90) | $0.33 | $2.1 | $0.03 | $0.77 | 99.95% |
DeepInfra | Nvidia H100 / L40S | 81 t/s | 2.75s (3.83s p90) | $0.23 | $1.5 | $0.025 | $0.55 | 99.90% |
Fireworks AI | Nvidia B200 / H100 | 118 t/s | 2.14s (2.91s p90) | $0.3 | $1.95 | $0.03 | $0.71 | 99.99% |