Mercury 2.5

All projects

Apps & Tools

A high-speed diffusion-based LLM running at over 1,100 tokens per second.

byInception

Playable

Mercury 2.5 media is blocked

Allow external media to connect to the provider and play this content.

Description

Mercury 2.5 is a production-ready diffusion large language model designed for extreme low-latency workloads like voice agents and search infrastructure. It achieves speeds of 1,107 tokens per second on NVIDIA GPUs while offering a 40% increase in intelligence over its predecessor, suitable for complex tasks like context compaction and reasoning.

Descriptions, tags, and model credits may be AI-generated or inferred from public sources and can be incomplete or wrong. Learn more.

Details

Why this is interesting: A rare production-scale diffusion LLM optimized for extreme speed, powering voice and search agents at over 1,100…
Attributed to: @_inception_ai
Built with: Mercury 2.5 Seedance 2.5 Grok Voice Think Fast 1.0 Veo 3.1 Fast Wan 2.5 Redwood AI Mercury Coder
Links: inceptionlabs.ai
View tweet
Tags: #llm #diffusion #real-time #api

Discovered: X
First posted: X · Sep 9, 2026
Listed: Added 7d ago
Build: Strong
Models: Strong

Mercury 2.5

Description

Timeline

Mercury 2.5

Description

Timeline