Keye-VL-1_5-8B
Will it run on your hardware?
Pick your GPU memory - see which quantizations fit, and the cheapest card for the rest
Need an exact figure for your context length? Use the VRAM calculator.
Run it locally
Copy-paste - running in under a minute
vllm serve Kwai-Keye/Keye-VL-1_5-8BNew to this? Start with Ollama · serve to many users with vLLM.
Deep dive
Notes, sources, and the full write-up
Keye-VL-1_5-8B is an open-weight video understanding model from Kwai-Keye, published on Hugging Face in August 2025. It has 8.7B parameters and is released under the apache-2.0 license.
HardwareLink to this section
At full FP16 precision the weights need about 17 GB of memory; 8-bit quantization brings that to roughly 9 GB. Check the VRAM tiers above to see which GPUs fit it.
PopularityLink to this section
Keye-VL-1_5-8B had 5,346 downloads on Hugging Face in the last month and has 72 likes.
Run it locallyLink to this section
The weights are on Hugging Face and load with Hugging Face Transformers. Follow the model card for the exact setup and license terms.
Frequently asked
Quick answers to common questions
How much VRAM does Keye-VL-1_5-8B need?
Keye-VL-1_5-8B with 8.7B parameters needs approximately 5 GB at Q4_K_M quantization. Use our VRAM calculator for an exact estimate.
Is Keye-VL-1_5-8B better than other Kwai-Keye models?
Keye-VL-1_5-8B has 8.7B parameters with 131,072 context - a strong choice for video-understanding, vision.
What license is Keye-VL-1_5-8B under?
Keye-VL-1_5-8B is released under the apache-2.0 license, making it suitable for most commercial and personal projects.
What hardware runs Keye-VL-1_5-8B well?
With 8.7B parameters, Keye-VL-1_5-8B requires adequate VRAM. High-end GPUs like the RTX 4090 (24GB), RTX 5090 (32GB), or Mac Studio with unified memory are good options. Check our hardware directory for specific recommendations.
What is the best quantization for Keye-VL-1_5-8B?
Q4_K_M is the recommended sweet spot - ~98% of FP16 quality at ~27% of the size. Q5_K_M (~6 GB) is an option if you have spare VRAM. Use our VRAM calculator to compare.
How long can Keye-VL-1_5-8B's context window handle?
Keye-VL-1_5-8B supports a 131,072-token context window - enough for very long documents, codebases, or multi-turn conversations. Real-world usable context may vary by implementation.
What models compete with Keye-VL-1_5-8B?
Keye-VL-1_5-8B competes with other models in its class. Browse our model directory for comparisons, benchmarks, and community reviews to find the best fit.
Nearby options
Similar models and compatible hardware by spec