SGLang supported models: Multimodal Language Models

Read 2026-10-03. Speech and audio rows confirmed: openai/whisper-large-v3 and Qwen/Qwen3-ASR-1.7B are served on the audio/transcriptions route; the page's own descriptor for the KDA/MLA hybrid class appears as 'hybrid KDA/MLA' in the model-family column. Vision rows include llava-onevision (lmms-lab/llava-onevision-qwen2-7b-ov), llava-next, LLaVA-NeXT-Video, Ling-3.0-flash-VL, LFM2.5-VL, MiniCPM-V-2, GLM-4.5V, GLM-OCR, Qwen3-VL 235B-A22B and 30B-A3B, Qwen3-Omni-30B-A3B, Kimi-VL-A3B-Instruct, MiMo-VL-7B-RL, InternLM2-VL (meta-llama/Llama-3.2-11B-Vision-Instruct), Nemotron Nano 12B v2 VL, Phi-4-multimodal, Mistral-Small-3.1-24B, Step3-VL-10B, dots.ocr, PaddleOCR-VL, Cosmos3, deepseek-vl2, Janus-Pro-7B, NVILA-8B, gemma-3-4b-it. NO encoder-decoder or pooling/embedding architecture list lives on this page; it is generative multimodal only.