What does your model need?
Find an open model, see the GPU memory it takes to run or fine-tune, and the cheapest GPU that fits.
652 models. Memory marked ~ is an estimate from the model's size. Prices are full price for one GPU in stock, without VAT.
| Model | Task | GPU memory | Cheapest GPU that fits | Parameters | Downloads | License |
|---|---|---|---|---|---|---|
| distilgpt2 distilbert | Text generation | ~0.2 GB | €0.04/h RTX 3060 | 0.1B | 1.5M | Apache 2.0 |
| Ornith-1.0-35B ornith-ai | Text generation | ~78 GB | €2.09/h RTX PRO 6000 | 35.1B | 1.5M | MIT |
| jina-embeddings-v3 jinaai | Feature extraction | ~1.3 GB | €0.04/h RTX 3060 | 0.6B | 1.5M | CC BY-NC 4.0 |
| Mistral-7B-Instruct-v0.2 mistralai | Text generation | ~16 GB | €0.17/h RTX 3090 | 7.2B | 1.5M | Apache 2.0 |
| Voxtral-Mini-4B-Realtime-2602 mistralai | Automatic speech recognition | — | — | 4.4B | 1.5M | Apache 2.0 |
| Qwen3.6-27B-FP8 Qwen | Image text to text | ~34 GB | €0.62/h RTX 6000 Ada | 27.8B | 1.4M | Apache 2.0 |
| SapBERT-from-PubMedBERT-fulltext cambridgeltl | Feature extraction | ~0.2 GB | €0.04/h RTX 3060 | 0.1B | 1.4M | Apache 2.0 |
| GLM-5 zai-org | Text generation | ~1685 GB | — | 754B | 1.4M | MIT |
| chandra-ocr-2 datalab-to | Image text to text | ~12 GB | €0.13/h RTX 5070 | 5.3B | 1.4M | OpenRAIL |
| Qwen2.5-3B-Instruct-AWQ Qwen | Text generation | ~7.6 GB | €0.04/h RTX 3060 | 3.4B | 1.4M | Other |
| wav2vec2-xls-r-300m-cs-250 comodoro | Automatic speech recognition | — | — | 0.3B | 1.4M | Apache 2.0 |
| gemma-4-26B-A4B-it-FP8-dynamic RedHatAI | Image text to text | ~32 GB | €0.62/h RTX 6000 Ada | 26.5B | 1.3M | Apache 2.0 |
| Llama-3.2-1B-Instruct meta-llama | Text generation | ~2.8 GB | €0.04/h RTX 3060 | 1.2B | 1.3M | Llama 3.2 |
| w2v-bert-2.0 facebook | Feature extraction | ~1.3 GB | €0.04/h RTX 3060 | 0.6B | 1.3M | MIT |
| Qwen2.5-VL-7B-Instruct-AWQ Qwen | Image text to text | ~19 GB | €0.17/h RTX 3090 | 8.3B | 1.3M | Apache 2.0 |
| wav2vec2-large-voxrex-swedish KBLab | Automatic speech recognition | — | — | 0.3B | 1.3M | cc0-1.0 |
| Qwen2.5-Coder-7B-Instruct Qwen | Text generation | ~17 GB | €0.17/h RTX 3090 | 7.6B | 1.3M | Apache 2.0 |
| RealVisXL_V5.0 SG161222 | Text to image | — | — | 2.6B | 1.3M | OpenRAIL++ |
| jina-embeddings-v2-small-en jinaai | Feature extraction | ~0.1 GB | €0.04/h RTX 3060 | 0B | 1.3M | Apache 2.0 |
| Qwen3.5-35B-A3B-FP8 Qwen | Image text to text | ~42 GB | €0.62/h RTX 6000 Ada | 36B | 1.3M | Apache 2.0 |
| GLM-4.5-Air zai-org | Text generation | ~247 GB | — | 110B | 1.3M | MIT |
| Qwen2.5-VL-7B-Instruct-NVFP4 nvidia | Text generation | ~8.1 GB | €0.09/h RTX 3080 | 5.4B | 1.2M | Other |
| Inkling-Small thinkingmachines | Image text to text | ~594 GB | — | 266B | 1.2M | Apache 2.0 |
| Llama-3.2-1B-Instruct unsloth | Text generation | ~2.8 GB | €0.04/h RTX 3060 | 1.2B | 1.2M | Llama 3.2 |
| MiniMax-M2.5 MiniMaxAI | Text generation | ~257 GB | — | 229B | 1.2M | Other |