Rocket - Think it. Type it. Launch it.
Mini toolsSeed 1.5 VL
By ByteDance
Model family: Seed
Seed1.5-VL is ByteDance’s flagship vision-language model, pairing a SeedViT 532M vision encoder with a 20B-active MoE LLM. It handles images and videos of arbitrary aspect ratio, does fine-grained grounding, OCR and visual puzzles, and powers GUI agents for control and gameplay, while matching top VLMs with much lower compute.
New
Multimodal
Gen 3
Overview
Compact vision-language foundation model from ByteDance Seed, combining a 532M vision encoder with a 20B-active MoE LLM to deliver strong image, video and GUI understanding and multimodal reasoning, with many SOTA results at low inference cost.
About ByteDance
ByteDance is a multinational technology company known for its content platforms, including TikTok and Douyin.
View Company ProfileTools using Seed 1.5 VL
Last updated: February 25, 2026
