AI Model2026
Turns images into structured visual data—scene descriptions, objects, text, masks, and spatial relations—on local devices. Its quantized model is about 22 MB, targeting phones, wearables, and robots.
AI Model2026
Provides GGUF quantizations of a refusal-ablated Gemma 4 31B instruction model, with options spanning file size and precision plus separate image-processing components.
AI Model2026
Provides parallel speculative-decoding drafts for Qwen3.8-27B. It passes predicted features between adjacent positions across layers, improving acceptance length and throughput in the reported SGLang benchmarks.
AI Model2026
Combines text-to-image generation and reference-guided editing in one model, producing images around 16 megapixels and accepting up to 16 reference images.