Diffusion LLMs: Inception's Mercury 3 as the flagship claim
Where it stands: Inception has announced Mercury 3, calling it the most capable diffusion-based large language model. No pricing, availability, or independent benchmark results are known yet. Watch whether it actually ships, how independent tests score it, and whether diffusion models prove a speed or cost advantage.
Latest
- 2026-10-19 Inception announces Mercury 3 (~3:00 PM ET — unite.ai), describing it as "the most capable diffusion-based large language model" — the record's first thread on diffusion architecture as a model-release category. Announced, not shipped: no pricing, availability, or independent benchmark detail in the reporting, and the capability framing is the vendor's own. Watch: weights/API access actually becoming available, Artificial Analysis-class independent scores vs the "most capable diffusion LLM" claim, and whether diffusion models keep claiming a speed/cost lane against text-stream frontier models.