ImageBench V1 —

192 evaluations across 6 categories

Benchmark V1 verdicts are produced by VLM judges and can contain mistakes. Treat PASS/FAIL labels as machine-assisted assessments, and inspect the images yourself. Learn more about the methodology.

Generation Details

Source-backed model context, size, cost, and request settings for this ImageBench V1 run.

local/sensenova-u1.5-8b-mot-base

Local

SenseNova-U1.5-8B-MoT is SenseTime OpenSenseNova's Apache-2.0 native unified multimodal model: one 8B checkpoint covers text-to-image, image editing and visual question answering. Built on a Qwen3-8B backbone with Mixture-of-Transformers generation branches and a pixel-space generator, it targets legible Chinese and English text, infographics, native 4K and complex instruction following. This entry runs the base model without the 8-step LoRA.

Maker: SenseTime (OpenSenseNova)
Family: SenseNova-U1.5
Model Size: 8B
high
Cost: local run; no API price
not_applicable
Run Target: gx10/sensenova-u1.5-8b-mot-base
Effective Request: response_format: b64_json · size: 1024x1024 · num_inference_steps: 50 · guidance_scale: 4 · seed: 7
Sources: sensenova/SenseNova-U1.5-8B-MoT model card