StoryFirst seen
The startup's new release swaps the standard text-generation architecture for a diffusion approach, promising much faster output at the same price.
Everything it touches
2 names 2 voices 2 source posts
The story’s neighborhood
Explore the graph, or choose a connection below.
Follow the evidence
This brief comes from Tech Twitter’s edited headlines. The posts below are the sources grouped with the story; statements in them belong to their authors.
Inception just dropped Mercury 3, a model 4-8x faster than Luna / Haiku for the same price and benchmark performance. In real world use, OpenRouter shows that it’s effectively 2-4x faster and 2-4x cheaper effective price. The time for diffusion language models is finally here!Read original on X
Inception launched Mercury 3, claiming 1,068 tokens per second on NVIDIA GPUs and a 1 million token context window.Read original on X
Inception launches Mercury 3 with big speed claims for diffusion models
Inception launches Mercury 3 with big speed claims for diffusion models
Shared connections
The collaboration could help China train advanced AI models without relying on Nvidia, redrawing the map of the global chip
A reported order worth more than $35 billion against allocations of just 5 to 10 percent shows Chinese AI demand far outstripping supply.
Huawei engineers working inside DeepSeek's offices point to China's accelerating push to train advanced models without relying on Nvidia.
The Apache 2.0 release runs on both Nvidia and Huawei hardware, handing developers the serving stack behind DeepSeek's own low serving costs.