# AI Latency Tracker

> Independent, provider-neutral latency and uptime of AI inference APIs (OpenAI, Anthropic, Google, Mistral, DeepSeek, xAI, OpenRouter), measured from multiple regions with distributed probes and updated automatically. Freshness and per-region measurement are the differentiator — figures here are measured, not scraped from pricing pages.

## Regions
- [Fastest AI API from Asia (Tokyo)](https://llmlatency.dev/region/ap-tokyo): edge + inference latency ranking, updated automatically
- [Fastest AI API from Europe (Germany)](https://llmlatency.dev/region/eu-hetzner): edge + inference latency ranking, updated automatically
- [Fastest AI API from South America (São Paulo)](https://llmlatency.dev/region/sa-east): edge + inference latency ranking, updated automatically
- [Fastest AI API from US (Central)](https://llmlatency.dev/region/us-central): edge + inference latency ranking, updated automatically

## Providers
- [Is the ai21 API down?](https://llmlatency.dev/provider/ai21): live status, recent outages, latency and uptime of ai21 across regions
- [Is the aleph-alpha API down?](https://llmlatency.dev/provider/aleph-alpha): live status, recent outages, latency and uptime of aleph-alpha across regions
- [Is the Claude API (Anthropic) down?](https://llmlatency.dev/provider/anthropic): live status, recent outages, latency and uptime of anthropic across regions
- [Is the baichuan API down?](https://llmlatency.dev/provider/baichuan): live status, recent outages, latency and uptime of baichuan across regions
- [Is the baseten API down?](https://llmlatency.dev/provider/baseten): live status, recent outages, latency and uptime of baseten across regions
- [Is the Cerebras API down?](https://llmlatency.dev/provider/cerebras): live status, recent outages, latency and uptime of cerebras across regions
- [Is the Cohere API down?](https://llmlatency.dev/provider/cohere): live status, recent outages, latency and uptime of cohere across regions
- [Is the deepinfra API down?](https://llmlatency.dev/provider/deepinfra): live status, recent outages, latency and uptime of deepinfra across regions
- [Is the DeepSeek API down?](https://llmlatency.dev/provider/deepseek): live status, recent outages, latency and uptime of deepseek across regions
- [Is the doubao API down?](https://llmlatency.dev/provider/doubao): live status, recent outages, latency and uptime of doubao across regions
- [Is the ernie API down?](https://llmlatency.dev/provider/ernie): live status, recent outages, latency and uptime of ernie across regions
- [Is the featherless API down?](https://llmlatency.dev/provider/featherless): live status, recent outages, latency and uptime of featherless across regions
- [Is the Fireworks AI API down?](https://llmlatency.dev/provider/fireworks): live status, recent outages, latency and uptime of fireworks across regions
- [Is the friendli API down?](https://llmlatency.dev/provider/friendli): live status, recent outages, latency and uptime of friendli across regions
- [Is the glm API down?](https://llmlatency.dev/provider/glm): live status, recent outages, latency and uptime of glm across regions
- [Is the Gemini API (Google) down?](https://llmlatency.dev/provider/google): live status, recent outages, latency and uptime of google across regions
- [Is the Groq API down?](https://llmlatency.dev/provider/groq): live status, recent outages, latency and uptime of groq across regions
- [Is the hunyuan API down?](https://llmlatency.dev/provider/hunyuan): live status, recent outages, latency and uptime of hunyuan across regions
- [Is the hyperbolic API down?](https://llmlatency.dev/provider/hyperbolic): live status, recent outages, latency and uptime of hyperbolic across regions
- [Is the iflytek API down?](https://llmlatency.dev/provider/iflytek): live status, recent outages, latency and uptime of iflytek across regions
- [Is the inference-net API down?](https://llmlatency.dev/provider/inference-net): live status, recent outages, latency and uptime of inference-net across regions
- [Is the kimi API down?](https://llmlatency.dev/provider/kimi): live status, recent outages, latency and uptime of kimi across regions
- [Is the meta-llama API down?](https://llmlatency.dev/provider/meta-llama): live status, recent outages, latency and uptime of meta-llama across regions
- [Is the minimax API down?](https://llmlatency.dev/provider/minimax): live status, recent outages, latency and uptime of minimax across regions
- [Is the Mistral API down?](https://llmlatency.dev/provider/mistral): live status, recent outages, latency and uptime of mistral across regions
- [Is the nebius API down?](https://llmlatency.dev/provider/nebius): live status, recent outages, latency and uptime of nebius across regions
- [Is the novita API down?](https://llmlatency.dev/provider/novita): live status, recent outages, latency and uptime of novita across regions
- [Is the nscale API down?](https://llmlatency.dev/provider/nscale): live status, recent outages, latency and uptime of nscale across regions
- [Is the OpenAI API down?](https://llmlatency.dev/provider/openai): live status, recent outages, latency and uptime of openai across regions
- [Is the OpenRouter API down?](https://llmlatency.dev/provider/openrouter): live status, recent outages, latency and uptime of openrouter across regions
- [Is the Perplexity API down?](https://llmlatency.dev/provider/perplexity): live status, recent outages, latency and uptime of perplexity across regions
- [Is the qwen API down?](https://llmlatency.dev/provider/qwen): live status, recent outages, latency and uptime of qwen across regions
- [Is the reka API down?](https://llmlatency.dev/provider/reka): live status, recent outages, latency and uptime of reka across regions
- [Is the replicate API down?](https://llmlatency.dev/provider/replicate): live status, recent outages, latency and uptime of replicate across regions
- [Is the SambaNova API down?](https://llmlatency.dev/provider/sambanova): live status, recent outages, latency and uptime of sambanova across regions
- [Is the sarvam API down?](https://llmlatency.dev/provider/sarvam): live status, recent outages, latency and uptime of sarvam across regions
- [Is the sensenova API down?](https://llmlatency.dev/provider/sensenova): live status, recent outages, latency and uptime of sensenova across regions
- [Is the siliconflow API down?](https://llmlatency.dev/provider/siliconflow): live status, recent outages, latency and uptime of siliconflow across regions
- [Is the stepfun API down?](https://llmlatency.dev/provider/stepfun): live status, recent outages, latency and uptime of stepfun across regions
- [Is the targon API down?](https://llmlatency.dev/provider/targon): live status, recent outages, latency and uptime of targon across regions
- [Is the Together AI API down?](https://llmlatency.dev/provider/together): live status, recent outages, latency and uptime of together across regions
- [Is the TypeSafe API down?](https://llmlatency.dev/provider/typesafe): live status, recent outages, latency and uptime of typesafe across regions
- [Is the upstage API down?](https://llmlatency.dev/provider/upstage): live status, recent outages, latency and uptime of upstage across regions
- [Is the writer API down?](https://llmlatency.dev/provider/writer): live status, recent outages, latency and uptime of writer across regions
- [Is the xAI API down?](https://llmlatency.dev/provider/xai): live status, recent outages, latency and uptime of xai across regions
- [Is the yi-01ai API down?](https://llmlatency.dev/provider/yi-01ai): live status, recent outages, latency and uptime of yi-01ai across regions

## Key answers
- [What is the fastest AI API?](https://llmlatency.dev/fastest-ai-api): measured answer — no single provider is fastest everywhere; leaders per region and how much the same provider varies by origin
- [Which AI provider has the best uptime?](https://llmlatency.dev/best-uptime): measured uptime per provider with the worst region shown next to the average, alongside latency
- [AI model migration guides](https://llmlatency.dev/migrate): every announced model retirement with its date, the provider-documented replacement, and days remaining — one page per model, each linking its official source
- [What does the same AI model cost at different hosts, and which answers fastest?](https://llmlatency.dev/price-vs-latency): per-host prices for multi-hosted models next to independently measured latency; the same weights can cost 10x more at one provider than another
- [Is Claude Code or Codex slower today?](https://llmlatency.dev/coding-agents): hourly measured generation speed, thinking tokens and time to first token of Claude Code (Claude Opus 5.5) and Codex CLI (GPT-6 Astra) on paid subscriptions, plus accuracy on a fixed hard question panel; each agent compared with its own last 7 days
- [Which AI APIs went down, and was it the provider or my network?](https://llmlatency.dev/incidents): outage log measured from 4 regions; a failure seen from 2+ regions at once is the provider, a single-region failure is not called an outage
- [Was an AI API down on a given day?](https://llmlatency.dev/incidents): one page per UTC day — measured outages next to what each provider's status page said
- [Outage & deprecation alerts](https://llmlatency.dev/alerts): request a conversation about incident feeds, model retirements and tailored delivery; no payment or automatic subscription; RSS and JSON already live
- [Sample weekly provider report](https://llmlatency.dev/pilot-sample): seven complete UTC days for five selected providers; separate network and model series; downloadable JSON; a free preview of a tailored data pilot

## About
- [About](https://llmlatency.dev/about): who runs the site (founder, location), how it is funded, independence from providers, contact
- [Methodology](https://llmlatency.dev/methodology): how latency is measured — direct multi-region probes, TTFB/TTFT, p50/p95, Speed Index
- [Glossary](https://llmlatency.dev/glossary): definitions of TTFB, TTFT, p50, p95, uptime, Speed Index
- [Provider comparisons](https://llmlatency.dev/compare/anthropic-vs-openai): head-to-head latency for popular pairs (e.g. openai vs anthropic, cerebras vs groq)
- [Data & licensing](https://llmlatency.dev/data): everything published is free under CC BY 4.0; bulk history, extra regions and providers on request
- [Which AI provider SDKs do developers actually install?](https://llmlatency.dev/adoption): monthly PyPI and npm download counts per provider SDK, ranked, with 30-day growth — reach, measured from the official statistics endpoints
- [Embeddable badges](https://llmlatency.dev/badge): self-updating SVG status (up / degraded / down) and measured latency for any provider
- [AI model deprecation & migration calendar](https://llmlatency.dev/deprecations): when models are retired and what to migrate to
- [Changelog](https://llmlatency.dev/changelog): most recent measurements and ranking changes
- [Corrections & disputes](https://llmlatency.dev/methodology#corrections): contact@llmlatency.dev — dispute a figure with provider, region and UTC timestamp; every number is backed by stored raw probes and confirmed errors are logged in the changelog (https://llmlatency.dev/changelog#corrections)