Mercury 3 is live through several API surfaces; latency and quality comparisons remain vendor-reported.
Inception launches Mercury 3 as a diffusion language model
2026-10-19Verified announcementTRIAL
Inception has launched Mercury 3, a diffusion language model available through its own OpenAI-compatible API, OpenRouter, Baseten and Vercel. The company offers 100 million free tokens for trials and reports lower latency with stronger evaluation and production results, but does not publish enough independent or same-condition evidence on the launch page to verify those comparisons.
Availability: Inception API, OpenRouter, Baseten and Vercel; OpenAI-compatible
What this means for you
For latency-sensitive agents, diffusion decoding is a materially different serving path and the live API makes a bounded trial possible. Compare end-to-end task success, output length, P95 latency and total cost on the same prompts before changing routing.
Evidence & related material
Technical details & scores
| Availability | Inception API, OpenRouter, Baseten and Vercel; OpenAI-compatible |
|---|---|
| Trial | 100 million free tokens advertised |
| Evidence Limit | latency, evaluation and production comparisons are vendor-reported |
| Time Basis | official page gives 2026-10-19 without a time; it was publicly linked by 2026-10-19T17:27:11Z, inside this run window |
Technical impact 4
Engineering value 4
Adoption signal 2
Maturity 4
Verification cost 2
Risk 3