LLM hosting by vendor
Every vendor compared: what it sells for running language models, from its public list prices, and where it runs.
| Vendor | Kind | Models sold per token | Kinds of GPU | Regions | Countries |
|---|---|---|---|---|---|
| Alibaba Cloud | Hyperscaler | 8 | 5 | 11 | 12 |
| Anthropic | Model maker | 13 | 0 | 1 | one price |
| AWS | Hyperscaler | 34 | 9 | 36 | 27 |
| Azure | Hyperscaler | 39 | 8 | 50 | 32 |
| Cloudflare | Edge network, Per-token API | 16 | 0 | 1 | one price |
| Core42 | Sovereign cloud | 22 | 4 | 4 | 3 |
| Crusoe | GPU cloud, Per-token API | 4 | 5 | 1 | one price |
| Databricks | Data platform | 43 | 0 | 1 | 1 |
| DeepInfra | Inference provider, Per-token API, GPU cloud | 36 | 5 | 1 | one price |
| Fireworks AI | Inference provider, Per-token API, GPU cloud | 6 | 5 | 2 | 1 |
| Gemini API | Model maker | 3 | 0 | 1 | one price |
| Google Cloud | Hyperscaler | 0 | 5 | 2 | 26 |
| Groq | Inference provider, Per-token API | 3 | 0 | 1 | one price |
| HF Endpoints | Model hub, Dedicated endpoints | 0 | 7 | 1 | one price |
| Huawei Cloud | Hyperscaler, Per-token API | 4 | 0 | 1 | one price |
| IBM watsonx | Enterprise AI platform | 3 | 4 | 1 | one price |
| Mistral | Model maker | 4 | 0 | 1 | one price |
| Nebius | GPU cloud, Per-token API | 17 | 6 | 9 | 6 |
| OCI | Hyperscaler | 6 | 11 | 11 | 23 |
| OpenAI | Model maker | 9 | 0 | 1 | one price |
| Snowflake | Data platform | 46 | 0 | 1 | one price |
| Telnyx | Inference API | 8 | 0 | 2 | 1 |
| Together AI | GPU cloud, Per-token API | 7 | 3 | 1 | one price |
| xAI | Model maker | 6 | 0 | 1 | one price |