AI Latency Tracker

Updated September 21, 2026 · 45 providers · 4 regions

As of September 21, 2026, the fastest-responding AI inference API by edge latency (time-to-first-byte) is fireworks at 22 ms p50, measured from Asia (Tokyo) (100% uptime, n=287). Rankings differ by region — see the table below.

Independent, provider-neutral latency and uptime of AI inference APIs (OpenAI, Anthropic, Google, Mistral, DeepSeek, xAI, OpenRouter and more), measured directly from multiple regions and updated automatically. How this is measured →

Is any AI API down right now?

all probes succeededsome probes failedhalf or more failedno data

Hour by hour for the last 72 hours, all regions combined; hover a cell for the count. Each provider page shows the same view per region. Full incident log →

Which AI API is fastest right now?

Which AI APIs are fastest overall?

Composite score 0–100 (100 = as fast as the region leader), averaged across all 4 measured region(s). How it’s computed →

Full ranking — Europe (Germany)

Edge latency (time-to-first-byte), p50/p95, last 24h. Other regions: see the per-region pages below.

How fast is each region?

• Fastest AI API from Asia (Tokyo)
• Fastest AI API from Europe (Germany)
• Fastest AI API from South America (São Paulo)
• Fastest AI API from US (Central)

How fast is a specific provider?

• ai21 latency by region
• aleph-alpha latency by region
• anthropic latency by region
• baichuan latency by region
• baseten latency by region
• cerebras latency by region
• cohere latency by region
• deepinfra latency by region
• deepseek latency by region
• doubao latency by region
• ernie latency by region
• featherless latency by region
• fireworks latency by region
• friendli latency by region
• glm latency by region
• google latency by region
• groq latency by region
• hunyuan latency by region
• hyperbolic latency by region
• iflytek latency by region
• inference-net latency by region
• kimi latency by region
• meta-llama latency by region
• minimax latency by region
• mistral latency by region
• nebius latency by region
• novita latency by region
• nscale latency by region
• openai latency by region
• openrouter latency by region
• perplexity latency by region
• qwen latency by region
• reka latency by region
• replicate latency by region
• sambanova latency by region
• sarvam latency by region
• sensenova latency by region
• siliconflow latency by region
• stepfun latency by region
• targon latency by region
• together latency by region
• upstage latency by region
• writer latency by region
• xai latency by region
• yi-01ai latency by region

Which provider is faster than which?

• anthropic vs openai latency
• anthropic vs google latency
• google vs openai latency
• cerebras vs groq latency
• deepseek vs openai latency
• groq vs sambanova latency
• fireworks vs together latency
• mistral vs openai latency