Qianfan-VL-8B
Qianfan-VL-8B combines an InternViT image encoder with Llama 3.1. It uses dynamic image patches for inputs up to 4K and has a documented 32K context.
Model design and access
The 8B release supports chain-of-thought responses. Its included license combines Baidu’s MIT terms with the Llama 3.1 conditions for the underlying language model.Qianfan-VL-8B official model card
Practical use
Practical guidance. Use this size to evaluate local multimodal reasoning before committing to the 70B model. Measure answer accuracy and memory at the required image resolution. Keep image preprocessing identical when comparing checkpoints.Qianfan-VL-8B official model card
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Context window | 32,768 tokens Documented checkpoint context window; includes input and generated text. Qianfan-VL-8B official model card |
| License | MIT terms for Baidu contributions plus Meta Llama 3.1 Community LicenseQianfan-VL-8B repository license |
| Language-model ancestry | Llama 3.1Qianfan-VL-8B official model card |
Related models
- Qianfan-VL-3B
Baidu · Qianfan-VL
Vision-language model combining Qwen2.5-3B with an InternViT encoder for images up to 4K.
Launch date not confirmedDownloadable - Qianfan-VL-70B
Baidu · Qianfan-VL
Seventy-billion-parameter Qianfan vision-language model built from Llama 3.1 and an InternViT image encoder.
Launch date not confirmedDownloadable