Free AI Hosting | Free.ai
Host AI models for free. GPU access, API hosting, and cloud deployment.
Aloxado na nube
Empregar a infraestrutura Free.ai. Cero configuración, cero mantemento. Todos os modelos están precargados e listos para usar mediante API ou interface web.
Dispoñible agoraDocker auto- hospedado
Execute os nosos modelos de IA de código aberto no seu propio hardware. Imaxes Docker con soporte para GPU, optimizadas para a inferencia.
AutoservizoPrivado xestionado
Servidores de GPU dedicados xestionados por nós, implantados na súa rexión de nube preferida. Illamento completo dos datos e SLA personalizado.
EnterpriseCity in California USADespliegue auto- hospedado
Todos os nosos modelos son de código aberto (Apache 2. 0 / MIT). Pode executalos na súa propia infraestrutura de GPU:
# Pull and run a model with Docker
docker pull ghcr.io/free-ai/inference:latest
docker run --gpus all -p 8000:8000 ghcr.io/free-ai/inference:latest \
--model qwen2.5-72b --quantization awqRequisitos mínimos
- GPU NVIDIA con 24GB+ VRAM (RTX 4090, A5000, A100)
- CUDA 12.0+ e Docker con NVIDIA Container Toolkit
- 16GB+ RAM do sistema, 100GB+ almacenamento por modelo
- Para modelos con parámetros 72B: 80 GB de VRAM (A100) ou configuración multi-GPU
Por que o auto-host?
- Privacidade dos datos — Your data never leaves your servers
- Sen límites de velocidade — Unlimited inference on your hardware
- Cumprimento — Meet data residency requirements
- Personalización — Fine-tune models on your data
- Control de custos — Fixed hardware costs, no per-token fees
- Air- gap — Runs fully offline
FAQ
Three options: Cloud Hosted (use our infrastructure, zero setup), Docker Self-Hosted (run models on your own GPU hardware), and Managed Private (dedicated GPU servers managed by us in your preferred region).
You need an NVIDIA GPU with 24GB+ VRAM (RTX 4090, A5000, A100), CUDA 12.0+, Docker with NVIDIA Container Toolkit, 16GB+ system RAM, and 100GB+ storage per model. For 72B parameter models, you need 80GB VRAM or a multi-GPU setup.
Yes. Self-hosted deployments run fully offline once the Docker images and model weights are downloaded. This is ideal for air-gapped environments and sensitive data processing.
Pull our Docker image and run it with GPU support. The command is: docker run --gpus all -p 8000:8000 ghcr.io/free-ai/inference:latest --model qwen2.5-72b --quantization awq. The container handles model loading and serves an API endpoint.
All self-hosted models use permissive open-source licenses -- Apache 2.0, MIT, or BSD. You can use them commercially without restrictions. We deliberately exclude models with restrictive licenses like Meta's Llama license.
Managed private hosting gives you dedicated GPU servers in your preferred cloud region, fully managed by our team. We handle setup, patching, model updates, and monitoring. You get full data isolation with an enterprise SLA.
Yes. Since all models are open-source, you can fine-tune them on your own data using standard training frameworks like Hugging Face Transformers. Our Docker images are compatible with popular fine-tuning tools.
Contact our sales team to discuss a trial period. We typically offer a short evaluation period for enterprise prospects to test managed private hosting before committing to a long-term plan.
Cloud hosting uses the standard token-based pricing. Self-hosted is free -- you only pay for your own hardware and electricity. Managed private hosting is priced based on GPU allocation, region, and SLA level.
Yes. You can self-host specific models for high-volume or sensitive workloads while using the Free.ai cloud for everything else. The API format is identical, making it easy to route requests between your infrastructure and ours.
We provide documentation, Docker images, and community support for self-hosted deployments. Managed private hosting includes full technical support, monitoring, and a dedicated account manager.
Cloud hosted is best for teams that want zero maintenance. Self-hosted is ideal for data privacy, compliance, or unlimited usage on your own hardware. Managed private is the best of both worlds -- full data isolation with no operational burden.