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 |
|---|---|---|---|---|---|---|
| Llama-3.2-3B-Instruct meta-llama | Text generation | ~7.2 GB | €0.04/h RTX 3060 | 3.2B | 872.2K | Llama 3.2 |
| LightOnOCR-2-1B lightonai | Image text to text | ~2.2 GB | €0.04/h RTX 3060 | 1B | 871.7K | Apache 2.0 |
| Qwen2.5-VL-3B-Instruct-AWQ Qwen | Image text to text | ~8.4 GB | €0.09/h RTX 3080 | 3.8B | 866.9K | — |
| Laguna-XS.2 poolside | Text generation | ~75 GB | €2.09/h RTX PRO 6000 | 33.4B | 860.1K | Apache 2.0 |
| SmolLM2-135M-Instruct HuggingFaceTB | Text generation | ~0.3 GB | €0.04/h RTX 3060 | 0.1B | 847.5K | Apache 2.0 |
| w2v-xls-r-uk Yehor | Automatic speech recognition | — | — | 0.3B | 843.6K | Apache 2.0 |
| Qwen3-TTS-12Hz-0.6B-Base Qwen | Text to speech | — | — | 0.9B | 841.5K | Apache 2.0 |
| Qwen3.5-397B-A17B Qwen | Image text to text | ~902 GB | — | 403B | 836.7K | Apache 2.0 |
| Laguna-M.1 poolside | Text generation | ~505 GB | — | 226B | 824.2K | Apache 2.0 |
| NVIDIA-Nemotron-3.5-Lightning-30B-A3B-FP8 RedHatAI | Text generation | ~36 GB | €0.62/h RTX 6000 Ada | 31.6B | 820.3K | — |
| parakeet-tdt-0.6b-v2 mlx-community | Automatic speech recognition | — | — | 0.6B | 814.6K | CC BY 4.0 |
| Ornith-1.0-9B ornith-ai | Text generation | ~21 GB | €0.17/h RTX 3090 | 9.4B | 814K | MIT |
| gemma-2-9b-it-AWQ-INT4 hugging-quants | Text generation | ~21 GB | €0.17/h RTX 3090 | 9.2B | 803.4K | Gemma |
| vllm-translategemma-4b-it Infomaniak-AI | Image text to text | ~9.6 GB | €0.09/h RTX 3080 | 4.3B | 799.8K | Gemma |
| Qwen3.5-0.8B-Base Qwen | Image text to text | ~2 GB | €0.04/h RTX 3060 | 0.9B | 797.9K | Apache 2.0 |
| Qwen3.6-35B-A3B-NVFP4-Fast unsloth | Image text to text | ~24 GB | €0.62/h RTX 6000 Ada | 19.8B | 794.6K | Apache 2.0 |
| nb-wav2vec2-1b-nynorsk NbAiLab | Automatic speech recognition | — | — | 1B | 775.5K | Apache 2.0 |
| Qwen2.5-32B-Instruct Qwen | Text generation | ~73 GB | €2.09/h RTX PRO 6000 | 32.8B | 775.4K | Apache 2.0 |
| gemma-3-27b-it-int4-awq gaunernst | Image text to text | ~61 GB | €2.09/h RTX PRO 6000 | 27.4B | 769K | Gemma |
| Qwen3.5-9B-MLX-4bit lmstudio-community | Image text to text | ~21 GB | €0.17/h RTX 3090 | 9.4B | 757.1K | Apache 2.0 |
| distil-large-v3 distil-whisper | Automatic speech recognition | — | — | 0.8B | 747.1K | MIT |
| Kimi-K3 moonshotai | Image text to text | ~3171 GB | — | 2780B | 742.4K | Other |
| Qwen3-TTS-12Hz-0.6B-CustomVoice Qwen | Text to speech | — | — | 0.9B | 734.1K | Apache 2.0 |
| VibeVoice-ASR microsoft | Automatic speech recognition | — | — | 8.7B | 714.5K | MIT |
| Qwen3.5-9B-MLX-8bit lmstudio-community | Image text to text | ~21 GB | €0.17/h RTX 3090 | 9.4B | 704.5K | Apache 2.0 |