Seed1.8 Model Card: Towards Generalized Real-World Agency

TL;DR

Seed1.8 is a versatile foundation model designed for complex, multi-step real-world tasks involving multi-turn interactions, tool use, and multi-modal understanding, with optimized deployment features.

Contribution

It introduces Seed1.8, a unified model supporting multi-turn interaction, tool use, and multi-modal tasks, with deployment optimizations for real-world applications.

Findings

Strong performance on standard benchmarks and workflows.

Supports multi-turn, multi-modal, and agentic behaviors.

Offers latency- and cost-aware inference modes.

Abstract

We present Seed1.8, a foundation model aimed at generalized real-world agency: going beyond single-turn prediction to multi-turn interaction, tool use, and multi-step execution. Seed1.8 keeps strong LLM and vision-language performance while supporting a unified agentic interface-search, code generation and execution, and GUI interaction. For deployment, it offers latency- and cost-aware inference, including configurable thinking modes and optimized visual encoding for images and video. We report evaluations on standard benchmarks and application-aligned workflows spanning foundational skills, multimodal understanding, and agentic behavior. Seed1.8 is released to support further research and development on interactive, real-world use cases.