Inception Debuts Mercury 3, Claims 4x-8x Agentic Edge — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Inception debuted Mercury 3, described in the article as a diffusion LLM, claiming top performance on agentic benchmarks against GPT-6 Luna, Claude Haiku 5.5, and GLM-5.3 Flash.
- Mercury 3 is stated to run 4x to 8x faster than the compared models, per Inception's own claims, framed by the article against an AI industry that has bet heavily on scaling LLMs and the data centers they require.
Why it matters: Inception positions Mercury 3 as an alternative to the AI industry's bet on large data centers for LLM scaling, claiming the diffusion model outperforms named competitors on agentic benchmarks at 4x-8x the speed. The performance numbers cited in the article come from Inception's own testing rather than independent verification.
Ask SkimNews
