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 |
|---|---|---|---|---|---|---|
| higgs-audio-v3-tts-4b-transformers multimodalart | Text to speech | — | — | 4.7B | 25.5K | Other |
| Xiaomi-OCR-0 xiaomi-research | Image text to text | ~2 GB | €0.04/h RTX 3060 | 0.9B | 25.1K | Apache 2.0 |
| SteadyDancer-14B MCG-NJU | Image to video | — | — | 16.4B | 25K | Apache 2.0 |
| Spark-X2.6-4B XHToken | Text generation | ~9.2 GB | €0.09/h RTX 3080 | 4.1B | 25K | Apache 2.0 |
| Wan2.2-I2V-A14B-Lightning-Diffusers magespace | Text to video | — | — | 14.3B | 24.9K | Apache 2.0 |
| VieNeu-TTS-v2 pnnbao-ump | Text to speech | — | — | 0.3B | 24.8K | Apache 2.0 |
| MiMo-V2.6-Pro-RL XiaomiMiMo | Text generation | ~1157 GB | — | 1024B | 24.3K | MIT |
| phonon-40m-fr gradium | Text to speech | — | — | 0.1B | 24.1K | CC BY-NC 4.0 |
| Qwen3-TTS-12Hz-1.7B-Base-bf16 mlx-community | Text to speech | — | — | 1.9B | 23.7K | Apache 2.0 |
| F2LLM-v2-80M codefuse-ai | Feature extraction | ~0.2 GB | €0.04/h RTX 3060 | 0.1B | 23.4K | Apache 2.0 |
| parler-tts-mini-v1 parler-tts | Text to speech | — | — | 0.9B | 21.8K | Apache 2.0 |
| MOSS-Image-SFT OpenMOSS-Team | Text to image | — | — | 5.1B | 21.5K | Apache 2.0 |
| MioTTS-2.6B Aratako | Text to speech | — | — | 2.6B | 21.2K | Other |
| Wan2.2-TI2V-5B-FP8 unsloth | Text to video | — | — | 5B | 20.9K | Apache 2.0 |
| Nanbeige4.3-33B Nanbeige | Text generation | ~74 GB | €2.09/h RTX PRO 6000 | 33.1B | 20.7K | Apache 2.0 |
| Dia-1.6B-0626 nari-labs | Text to speech | — | — | 1.6B | 20.6K | Apache 2.0 |
| Wan2.1-VACE-1.3B-diffusers Wan-AI | Image to video | — | — | 2.2B | 20.5K | Apache 2.0 |
| Sage-A1.5 InternScience | Text generation | ~80 GB | €2.09/h RTX PRO 6000 | 36B | 20.5K | Apache 2.0 |
| Nemotron-3.5-Parse nvidia | Image text to text | ~1.4 GB | €0.04/h RTX 3060 | 0.6B | 20.3K | openmdw-1.1 |
| moonshine-tiny moonshine-ai | Automatic speech recognition | — | — | 0B | 20.3K | MIT |
| MOSS-TTS-Realtime OpenMOSS-Team | Text to speech | — | — | 2.3B | 19.6K | Apache 2.0 |
| LongLive-2.0-5B-Diffusers Rabinovich | Text to video | — | — | 5B | 19.5K | — |
| DeepSeek-V4.1-Flash-UNCENSORED-FP8 dealignai | Image text to text | ~855 GB | — | 763B | 19.4K | MIT |
| Cosmos3-Super-Image2Video nvidia | Image to video | — | — | 64.6B | 19.2K | Other |
| phonon-40m-multilingual gradium | Text to speech | — | — | 0.1B | 18K | CC BY-NC 4.0 |