ByteDance Seed: Seed 1.6 Flash

Seed 1.6 Flash is a multimodal model from ByteDance that accepts text, images, and video as input and returns up to 32,768 tokens per response. Its 262,144-token context window is large enough for long documents or extended conversations, and it supports tool use and reasoning, making it usable for agentic workflows. Structured output support is unconfirmed based on available specifications. At $0.075 per million input tokens and $0.30 per million output tokens, Seed 1.6 Flash sits at the lower end of multimodal pricing, which makes it worth considering for cost-sensitive applications that require video understanding or long-context processing. The trade-off is transparency: there is currently no independent benchmark coverage, so its actual performance relative to competing models is unproven. Buyers who need verified quality benchmarks before committing should treat this model as experimental until third-party evaluations become available.