# FastInfra > FastInfra is a unified AI inference platform: one OpenAI-compatible API for 774+ models across multiple wholesale providers, plus GPU Cloud, CPU Cloud, and dedicated bare-metal servers — automatic cheapest-provider routing and transparent pricing. API base URL: `https://api.fastinfra.ai/v1` (drop-in replacement for the OpenAI SDK — change `base_url` and the API key only). Free tier: 12 models at $0. ## Docs & API - [API documentation](https://www.fastinfra.ai/Docs): authentication, chat completions, LTX video, Kokoro TTS, rate limits - [OpenAPI specification](https://www.fastinfra.ai/openapi.json): machine-readable API description - [Model catalog with live pricing](https://www.fastinfra.ai/Pricing): searchable per-token prices - [Model landing pages](https://www.fastinfra.ai/models): per-model pricing, providers, FAQs, code snippets - [Provider pages](https://www.fastinfra.ai/providers): upstream provider coverage - [AI crawler summary](https://www.fastinfra.ai/ai.txt): condensed discovery file ## Infrastructure products - [GPU Cloud configurator](https://gpu.fastinfra.ai/Gpu): rent H100/H200/A100 pods by the minute - [CPU Cloud configurator](https://cpu.fastinfra.ai/Cpu): Linux VMs with per-minute billing - [Dedicated servers](https://cpu.fastinfra.ai/Dedicated): bare-metal EX/AX/RX/SX/GPU lines and auction stock ## Solution guides - [OpenAI-Compatible API for Every Model](https://www.fastinfra.ai/solutions/openai-compatible-api) - [Cheapest LLM API — Per-Token Pricing](https://www.fastinfra.ai/solutions/cheapest-llm-api) - [GPU Cloud — Rent H100, H200, A100 Pods](https://www.fastinfra.ai/solutions/gpu-cloud-hosting) - [Cloud VPS — On-Demand Virtual Servers](https://www.fastinfra.ai/solutions/cloud-vps-hosting) - [Dedicated Servers — Bare-Metal Hosting](https://www.fastinfra.ai/solutions/dedicated-servers) - [AI Video Generation API — LTX](https://www.fastinfra.ai/solutions/ai-video-api) ## Full model list - [llms-full.txt](https://www.fastinfra.ai/llms-full.txt): every model with current USD prices per 1M tokens