DiffusionGemma
No written verdict yet. Here is what commenters said.
5recommend
0warn against
Jun 10, 2026first mention
Sep 27, 2026latest mention
Why people recommend it
“It (DiffusionGemma with the Jev mode) can also solve an ASCII maze.”
runs it on a DGX Sparkon “Reverse-engineered Jev-like model”
“I like the model a lot and it's fairly good at reasoning.”
Show 3 more
“Probably the fastest model I've been able to run so far”
“What makes me hopeful about this release is that possibly this same conversion can be applied”
“DiffusionGemma seem to both be very high token rate”
Caveats and dissent
No one in these comments warned against it.
Often compared with
Who recommends it in their own words
Threads where it came up
- DiffusionGemma Technical Report 3 mentions
- Reverse-engineered Jev-like model 2 mentions
- >1B tokens/minute/GPU by combining query planner and inference engine 1 mention
- OpenJev 1 mention
- Controlling Reasoning Effort in LLMs 1 mention
- Running local models is good now 1 mention
Data: diffusiongemma.json. Quotes are verbatim; the verdict and theme labels are written by a model from these comments, and every clause links to its sources.
All comments 11
“We have some more results on DiffusionGemma”
“DG wins some, Jev wins some, both show low confidence when wrong”
“DiffusionGemma has a vision tower and significantly more spatial data in its training set”
“It (DiffusionGemma with the Jev mode) can also solve an ASCII maze.”
“I like the model a lot and it's fairly good at reasoning.”
“Probably the fastest model I've been able to run so far”
“What makes me hopeful about this release is that possibly this same conversion can be applied”
“gonna give it a try with DiffusionGemma”
“DiffusionGemma seem to both be very high token rate”
“Algorithmic advances like DiffusionGemma ~4x token gen speeds”
“NVIDIA are hosting a free endpoint for this one”