Ovis2.5 9B
Ovis2.5 9B is an open-weights language model from Qwen, released August 15, 2025. It has 9.2B parameters (dense). The context window is 40K tokens.
We present Ovis2.5, a successor to Ovis2 designed for native-resolution visual perception and strong multimodal reasoning. Ovis2.5 integrates a native-resolution vision transformer that processes images at their native, variable resolutions, avoiding the degradation from fixed-resolution tiling and preserving both …
Check it against your own PC →
Can you run Ovis2.5 9B locally?
The smallest common GPU it runs on is 6 GB of VRAM (with 32 GB of RAM): Q6_K, split GPU / CPU, roughly 5 tokens per second. Estimates assume 32 GB of DDR5 RAM, a 16K context and llama.cpp-style GGUF quantization. Enter your exact hardware in the live app.
| GPU memory | Verdict | Best quant | VRAM | RAM | tok/s |
|---|---|---|---|---|---|
| 6 GB VRAM | Runs, slowly | Q6_K (≈ lossless) | 5.2 GB | 5.3 GB | ~5 |
| 8 GB VRAM | Runs, slowly | Q6_K (≈ lossless) | 7.2 GB | 3.3 GB | ~7 |
| 12 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 10.5 GB | – | ~28 |
| 16 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 10.5 GB | – | ~38 |
| 24 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 10.5 GB | – | ~69 |
| 32 GB VRAM | Runs on GPU | Q6_K (≈ lossless) | 10.5 GB | – | ~130 |
Benchmark results
Epoch AI has not published results for Ovis2.5 9B yet. New models are usually evaluated within days to weeks.
Links
Announcement · Hugging Face: ATH-MaaS/Ovis2.5-9B
More from Qwen
Qwen3.8 Omni Flash · Qwen3.8 Max (0902) · Qwen3.8 Flash · Qwen3.8 Flash Next · Qwen 3.8 27B · Qwen3.8-2.4T-A95B · Qwen 3.8 Max · Qwen3.7 Flash