Inference hosts
Serve open-weights models on their own GPUs, priced per token.
31 providers
| Provider | Categories | Owner | Pricing at a glance |
|---|---|---|---|
| AkashML | Inference hosts | (private) — built on Akash Network | 6 models |
| Atlas Cloud | Inference hosts | Atlas Cloud (private) | 29 models |
| Baseten | GPU clouds, Inference hosts | (private) | 13 models |
| Cerebras Inference | Inference hosts | Cerebras Systems Inc. (private) | 2 models |
| Chutes | Inference hosts | (private) — built on Bittensor subnet 64 (Rayon Labs) | 14 models |
| Cloudflare Workers AI | Inference hosts | Cloudflare, Inc. (NYSE: NET) | 28 models |
| CoreWeave | GPU clouds, Inference hosts | CoreWeave, Inc. (Nasdaq: CRWV) — owns Weights & Biases | 25 models · GPUs from $1.25/GPU-hr |
| Crusoe | GPU clouds, Inference hosts | Crusoe Energy Systems (private) | 12 models · GPUs from $1.50/GPU-hr |
| DeepInfra | Inference hosts | (private) | 48 models |
| DigitalOcean | Cloud VPS, GPU clouds, Inference hosts | DigitalOcean Holdings (NYSE: DOCN) | plans from $4/mo · 23 models · GPUs from $1.57/GPU-hr |
| Featherless | Inference hosts | (private) — Recursal AI | 45 models |
| Fireworks AI | Inference hosts | Fireworks AI, Inc. (private) | 16 models |
| FriendliAI | Inference hosts | FriendliAI Corp. (private) | 7 models |
| GMI Cloud Inference Engine | Inference hosts | GMI Cloud (private) | 32 models |
| Groq | Inference hosts | Groq, Inc. (private) | 7 models |
| Hyperbolic | Inference hosts | (private) | 2 models |
| Inference.net | Inference hosts | Inference R&D, Inc. (private) | 22 models |
| Lambda | GPU clouds, Inference hosts | Lambda, Inc. (private) | GPUs from $1.29/GPU-hr |
| Modal | Serverless containers, GPU clouds, Inference hosts | Modal Labs (private) | Serverless GPU per second: B300 $0.001972/s, B200 $0.001736/s, H200 $0.001261/s, H100 $0.… |
| Nebius | GPU clouds, Inference hosts | Nebius Group N.V. (NASDAQ: NBIS) | 7 models · GPUs from $1.55/GPU-hr |
| Novita AI | Inference hosts | (private) | 52 models |
| Nscale Serverless Inference | Inference hosts | Nscale Global Holdings (private) | 11 models |
| NVIDIA NIM (build.nvidia.com) | Inference hosts | NVIDIA Corp. (NASDAQ: NVDA) | Hosted NIM API endpoints free for prototyping to NVIDIA Developer Program members; produc… |
| OVHcloud AI Endpoints | Inference hosts | OVH Groupe SA (EPA: OVH) | 8 models |
| Parasail | Inference hosts | (private) | 36 models |
| Replicate | GPU clouds, Inference hosts | 13 models | |
| SambaNova Cloud (SambaCloud) | Inference hosts | SambaNova Systems, Inc. (private) | 7 models |
| Scaleway | Cloud VPS, GPU clouds, Inference hosts | Iliad Group (private; Scaleway SAS) | plans from $0.51/mo · 13 models · GPUs from $0.904/GPU-hr |
| SiliconFlow | Inference hosts | SiliconFlow (private) | 32 models |
| Together AI | GPU clouds, Inference hosts | Together Computer, Inc. (private) | 27 models · GPUs from $3.99/GPU-hr |
| Venice.ai API | Inference hosts | Venice.ai (private) | 33 models |