MIRA
AI
world model
SHOW HN2026
How HN took it
4 positive, 4 neutral or asking, 0 mixed, 1 critical, out of 9 comments about the project. Thread score +0.16.
Idea: 2 for, 1 againstPerformance: 2 for, 0 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Impressive coherence and near-infinite rollouts; feels like the real game
From 3 comments
I'm insanely impressed with MIRA! I mean WOW, I've been doing a lot of diffusion world model experiments and I had thought of trying stuff like "visual anchors" to support some…
blackfireWM on HNWow! At first, I expected this to be a demonstration of an AI playing rocket league, but I rapidly realized this is actually a model simulating rocket league. Wild! It feels just…
danking00 on HNVáclav here from the team, we're happy to answer questions :) The most surprising part to me is the auto-recovery behavior we mention at the end of the blog post, since any other…
vvolhejn on HN
Auto-recovery after going out of distribution is surprising
From 1 comment
Václav here from the team, we're happy to answer questions :) The most surprising part to me is the auto-recovery behavior we mention at the end of the blog post, since any other…
vvolhejn on HN
Multiplayer four-view input removes need to model other players
From 2 comments
A lot actually, since the model has all information given to it in the four views, it doesn't have to deal with any "theory of mind" of modeling the other players or being…
vvolhejn on HNthis is insane - what’s your thinking on how this improves model grounding and efficiency vs single pov outputs?
exortaz on HN
Objections
Not as a replacement for playing the game
From 1 comment
it would have been easier to just go to FNAC and buy Rocket League like a normal person :)
_willmanning on HN
Requests
Information on training compute
From 1 comment
How much compute did it take to train the model?
lostmsu on HN
Questions
Whether the model learned deeper world structure or just the training distribution
From 1 comment
Great work! My question is about what you mentioned at the end - how well do world models operate when out of distribution? In some sense we hope these models learn something…
shay_ker on HN
How multiplayer improves grounding versus single-view outputs
From 1 comment
this is insane - what’s your thinking on how this improves model grounding and efficiency vs single pov outputs?
exortaz on HN
Alternatives named
Playing Rocket League in the actual game
From 2 comments
it would have been easier to just go to FNAC and buy Rocket League like a normal person :)
_willmanning on HNIs this now the easiest way to play Rocket League on Linux?
LorenDB on HN