A new AI model writes its answers all at once, not word by word.
Startup Inception released Mercury 3 on Monday, a model that drafts whole answers in parallel instead of word by word. The company says it matches GPT-6 Luna and Claude Haiku 5.5 while running four to eight times faster. It is on sale now at $0.20 per million input tokens, and half that during launch.
Why it mattersFast answers matter most where people wait, like voice assistants, tutoring apps and support chats. If this method keeps scaling, speed may come from new model designs rather than only bigger chips.
Read in the app — free, in 9 languages
Related stories
MiniMax revealed that the mystery model Space Bunny was its own.2026-10-20DeepSeek quietly made its cheap open model a far better coder.
2026-10-17Nvidia squeezed an agent model into a gaming card's 24GB of memory.
2026-10-17Reflection's 501-billion-parameter model is now free to download.
2026-10-16Black Forest Labs' new open image model fits on an 8GB graphics card.
2026-10-16