Qwen3-VL-8B
A multimodal vision-language model in the Qwen series with 8B parameters, supporting joint understanding of visual and textual inputs. It is applied to tasks such as dual-view reasoning over time series data.
Timeline (last 90 days)
Qwen3-VL-8B improves by 11.2 points using OpenVLHarness.
The dual-view model in the paper is built on Qwen3-VL-8B.
On FastBench, Qwen3-VL-8B improves from 32.9% at 2 FPS to 44.6% at 24 FPS, but as history is compressed, the gains gradually saturate.
Loading the Qwen3-VL-8B model alone occupies approximately 17–18 GB of VRAM.
All Facts (4)
Qwen3-VL-8B improves by 11.2 points using OpenVLHarness.
50%UnverifiedThe dual-view model in the paper is built on Qwen3-VL-8B.
50%UnverifiedOn FastBench, Qwen3-VL-8B improves from 32.9% at 2 FPS to 44.6% at 24 FPS, but as history is compressed, the gains gradually saturate.
50%UnverifiedLoading the Qwen3-VL-8B model alone occupies approximately 17–18 GB of VRAM.
50%