SenseNova Open Sources U1.5 Lite: 8B Unified Multimodal Model with 4K Output
智东西 · wechat · 2026-08-21
SenseNova open-sourced the official version of SenseNova U1.5 Lite, a lightweight native unified multimodal large model. With only 8B parameters, it natively supports ultra-long complex instructions (3-4k characters) and 4K high-resolution image output. Based on the NEO-Unify architecture, U1.5 Lite fuses understanding, reasoning, generation, and editing capabilities into a single model using Multi-Expert Online Policy Distillation (MOPD), eliminating the need for explicit Router switching and reducing inference latency. Tests show the model accurately generates posters from prompts over 2000 characters with strong text rendering. It outperforms Qwen-Image-2.0, Seedream-4.5, and GPT-Image-1.5 on various benchmarks. It is now available on GitHub, HuggingFace, and ModelScope.
More from Models
- Peter Yang: Opus 5.5 Still Struggles to Hold Its Ground When Pushed Back — petergyang · 2026-10-16
- Robot Startup Ditches VLAs for Frontier LLMs in a Customer Deployment — ihorbeaver · 2026-10-16
- Musk amplifies user claim that Grok's computer use is fast enough to skip a Codex subscription — elonmusk · 2026-10-16
- PhoneLLM launches open voice-AI models with sub-100ms latency, tops PhoneBench at 83% — charles_irl · 2026-10-16
- Transluce vets: a year of neuron autointerp work still can't explain complex behaviors — aryaman2020 · 2026-10-16
- Grok Bot Ran a 12-Hour Data Analysis Nonstop, Handling API Rate Limits Itself — Daniel_Farinax · 2026-10-16