@FiniYang: Made a little cat survival game Had the official Jev and local Laya (mlx) team up to protect the kitten Each step chose…
Summary
The author created a cat survival game to test the performance of AI models Jev and Laya, finding that Jev is more accurate but slower, while Laya is fast but inaccurate, and plans to develop a leaderboard for such models.
View Cached Full Text
Cached at: 09/22/26, 07:56 PM
Made a little cat survival game
Had the official Jev and local Laya (mlx) team up to protect the kitten
Each step chosen by the model itself
The kitten would thus eat its fill, get injured, or go hungry
Conclusion: Jev is more accurate but slow, Laya is super fast but inaccurate
Tested each of the three levels three times
Official Jev cleared 3/9 rounds
All from the foraging level;
Local Laya cleared 0/9 rounds
Laya managed to return home every round, yet failed to meet the satiety, health, or warmth requirements for each level
Brothers, is the level design a bit too tough?
Planning to make a Jev-like decision model leaderboard for clearing levels later
Similar Articles
@0xCheshire: Jev is one of the best current System 1 decision-making models, and that's undeniable. But it has a fatal flaw: it's no…
The post compares AI models Jev and Laya, highlighting Laya's open-source, local, and low-latency capabilities for real-time decision-making in games.
@Yangtze_Seventh: Jev has only been blowing up for a few days, and its open-source rival Laya has already emerged. Neither generates long…
An open-source AI model named Laya has quickly emerged to rival the closed-source Jev, outperforming it in some metrics after fine-tuning, underscoring the rapid progress in open-source development versus secretive startup efforts.
I gave Jev, Laya, finetuned ModernCE and Qwen3.5 the controls to Doom
An experiment where AI models including Jev, Laya, finetuned ModernCE-base-nli, and Qwen3.5-4B are given control in the game Doom, with performance metrics measured for kill counts and survival time in different scenarios.
@AI_Jasonyu: Guys, you know I've been working on tool applications and have rarely touched games. But these past couple days, I felt like making a game called "Meow Tide". The basic gameplay is that hundreds of cats rush from all directions to rub against you, and when your stamina hits zero, you're "overwhelmed by cats", with the screen filling up with cat faces. Lasting 3 minutes until dawn counts as winning. I played for half an hour, and my best...
The author shares their experience of using the AI tool Gear Zero to create the game "Meow Tide". This tool can automatically generate game code, images, and sound effects from natural language descriptions.
@NFT_Chen: What?! A mere 0.6B local open-source decision model outperforms Laya on Typed Decisions! AgentJev-0.6B vs Laya: Accurac…
AgentJev-0.6B is a small open-source AI decision model that outperforms Laya on Typed Decisions with improved accuracy and significant computation reduction, enabling faster local inference.