Mercury 3
Use Mercury 3 for Fast Sales Follow-Ups Without Dropping Constraints
Set up Inception's diffusion text model Mercury 3 at Medium reasoning effort to turn call notes into accurate follow-up drafts.
- Beginner
- 3 steps
- Free (Public demo)
By TWK Guides DeskUpdated 18 Nov 2026
Every step checked against Inception's own docs on 18 Nov 2026
Overview
Inception's Mercury 3 uses parallel diffusion decoding to generate up to 1,068 tokens per second, making it well suited to repetitive tasks like turning call notes into follow-up drafts.
Before you start
- Access to Mercury Chat or an OpenRouter account
- Approved product pricing and plan rules to paste into the prompt
Step by step
0 of 3 done
Your ticks stay in this browser.
Step 1 of 3
Open Mercury Chat and set reasoning to Medium
The Mercury Chat interface defaults to Medium reasoning effort while the direct API defaults to Low, so verify the visible setting first.
Do this
0 of 3 done
What you should see
The composer displays Mercury 3 with Medium reasoning and web search disabled.
If it didn't work
If Medium is not selected, switch it manually before sending your first prompt.
Checked againstdocs.inceptionlabs.ai
Frequently asked questions
Can Mercury 3 process images or audio directly?
No. Mercury 3 is a text-only diffusion model, though it can be paired with separate speech tools for voice workflows.
