xAIAPI Host Benchmarks
Grok 4.6 (medium) — API Providers & Speed Matrix
Compare live throughput, TTFT latency distributions, prompt cache pricing, and hardware stacks across cloud hosting endpoints.
Fastest Provider
738 t/s
Lowest Latency (TTFT)
0.09s P50
Best Price Value
$0.9 / $3
Blended Pricing Workload Calculator
Switch between standard conversational ratios, agentic caching loops, and heavy RAG workloads.
| Provider & Host | Hardware Stack | Output Speed (TPS) | Latency TTFT (P50) | Input Price ($/1M) | Output Price ($/1M) | Cache Read ($/1M) | Blended Price ($/1M) | SLA |
|---|---|---|---|---|---|---|---|---|
xAIOfficial | Cloud TPU / Nvidia H100 | 59 t/s | 38.13s (53.38s p90) | $2 | $6 | $0.2 | $3 | 99.99% |
Groq Cloud | Groq LPU v2 (SRAM) | 248 t/s | 0.09s (0.12s p90) | $1.7 | $5.1 | $0.16 | $2.55 | 99.95% |
Cerebras Cloud | CS-3 Wafer-Scale Engine | 738 t/s | 0.12s (0.16s p90) | $1.8 | $5.4 | — | $2.7 | 99.90% |
Together AI | Nvidia H100 SXM5 | 89 t/s | 30.5s (41.94s p90) | $1.3 | $4.2 | $0.12 | $2.03 | 99.95% |
DeepInfra | Nvidia H100 / L40S | 77 t/s | 34.32s (47.66s p90) | $0.9 | $3 | $0.1 | $1.43 | 99.90% |
Fireworks AI | Nvidia B200 / H100 | 112 t/s | 26.69s (36.22s p90) | $1.2 | $3.9 | $0.12 | $1.88 | 99.99% |