Watch 14-Byte AI "brains" attempt to solve a 2D maze (Its hard)
Details
- External ID
- 49066083
- Source
- HN
- Company
- —
- Product
- Watch 14-Byte AI "brains" attempt to solve a 2D maze (Its hard)
- Website domain
- github.io
- Launched
- July 27, 2026
- Cohort
- —
- Upvotes
- 25
- Upvotes percentile
- 0.7592592592592593
- Tags
- —
- Fetched at
- Sept. 7, 2026, 9:26 p.m.
- Updated at
- Sept. 7, 2026, 9:26 p.m.
Description
Hey HackerNews,I built this project over the last few weeks as a palette cleanser from a failed game launch.I wanted to learn a bit about AI/Neural-Networks and naively thought I could build a tiny maze-solving AI in a weekend with a 100% solve rate.Well - I couldn't, but I got pretty close. 14 Bytes total model size, and a 96.5% solve rate on unseen mazes. Trained across 46 phases experimenting with different ideas to improve the model (better performance, smaller size).Its quite fun to watch the model attempt to solve the maze, when they fail its usually due to getting stuck in a loop. The models have no access to coordinates, map-data, or external memory scratches - they must navigate using only immediate local neighbourhood observations.There is a model dropdown and you can see how the model has progressed over each phase, constantly getting smaller and increasing its solve rate. Total trained models number in the thousands - I just expose the winning models from each phase.Overall a fun experiment, with much implementation help from AI agents to scaffold and implement the code (I'm a lazy software dev).
Enrichment
- Theme
- lightweight and on-device AI runtimes
- Vertical
- Horizontal
- Function
- Observability & eval
- Audience
- Developer
- AI stance
- AI feature
- Project type
- Hobby / open-source project
- Normalized one-liner
- watch small ai models solve mazes
- Manually corrected
- False
Could you build this?
Partial The visualization and maze simulation in a browser canvas is straightforward, but optimizing a functional neural network compressed into exactly 14 bytes requires specialized genetic algorithm or bit-packing math.
What it would actually take: The architecture comprises an interactive HTML5 canvas renderer and a simulation loop that executes micro-neural-network policies. The hard part is algorithmic: packing weights, activations, and routing into a 14-byte memory footprint while still maintaining valid pathfinding behaviors, typically solved via genetic evolution algorithms or constrained reinforcement learning. Implementing this requires knowledge of low-level data compression, evolutionary computation, and hyper-parameter tuning.
Discussion
9 comments analyzed.
Competitors mentioned: NEAT (Neuroevolution of Augmenting Topologies), MarI/O
Concerns raised: 14 byte metric doesn't include runtime, only weights, Experimentation outpaced understanding - claims need verification, Input information varies significantly across 2000+ trained models, Private repo - only frontend visible, not actual implementation
Feature requests: Add explicit (i) icon for tooltip areas on mobile/non-hover devices, Update frontend with legend and better visual alignment, Use LLM to auto-label model inputs instead of manual process, Longer postmortem/blog post explaining compression and discoveries
Competitors
Other products that read as similar to this one — 183 launches clear the similarity bar, closest 8 shown.
Attention rank: #59 of 184 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 269 days after the earliest competitor.
- I trained a chess engine to play like humans · hn · 2026-05-10 · 14 upvotes · similarity 0.47
- A puzzle I couldn't solve for a year · hn · 2026-08-21 · 18 upvotes · similarity 0.47
- Pencil Puzzle Bench · hn · 2026-03-03 · 5 upvotes · similarity 0.46
- Mini-AGI · hn · 2026-09-21 · 277 upvotes · similarity 0.43
- AI Bingo · hn · 2025-11-06 · 8 upvotes · similarity 0.43
- Solving Sudoku reasoning via Energy Geometric models · hn · 2026-02-12 · 18 upvotes · similarity 0.42
- AI Roundtable · hn · 2026-03-24 · 118 upvotes · similarity 0.42
- I successfully failed at one-shot-ing a video codec like h.264 · hn · 2026-04-04 · 9 upvotes · similarity 0.41
Other launches for this product
- No other launches for this product.
Same idea, different domain
Nobody's really built a observability & eval tool for Media & entertainment yet.