Midjourney is testing Thinking Mode for image generation in Alpha
The new rerun option gives the model more time to reason, aiming to improve prompt accuracy, typography, and visual coherence.
AT A GLANCE
Key points
- Available as Rerun (Thinking) on the Alpha website
- Targets prompt accuracy, typography, and coherence
- Runtime and credit cost have not been disclosed
- Broad availability is not yet confirmed
What is changing?
Midjourney has begun testing a new “Thinking Mode” for image generation on its web-based Alpha interface. According to the official October 8 announcement, users can run it by selecting “Rerun (Thinking)” in the lower-right corner of a completed job’s lightbox.
The company says its tests show improvements in prompt accuracy, typography, and overall visual coherence. Midjourney has not published comparative metrics, added runtime, or credit cost, however; this is an Alpha experiment collecting feedback rather than a broadly released product feature.
The announcement says the capability could later become a standalone thinking mode or an option to add more thinking to an existing generation. Users are being asked to share examples of where it helps or fails in the community channel.
The test suggests that image models may increasingly offer inference modes that spend longer evaluating prompts and composition instead of relying on a single pass.
Explore the original sources
Source date: 8 October 2026.
Published by AI Radar: 9 Oct 2026, 23:09 (Istanbul, UTC+3).
FOLLOW THE TOPIC
Topics and tags
Share your discovery.
Send this story to a friend.
Reader comments
What do you think about this development? Share your experience or question. Comments are published after editorial approval.
Your name/nickname and comment become public if approved. Read: Privacy notice.
Loading comments…