Jev Plays Pokémon

AI

game-playing agent

SHOW HN2026

282points
124comments
#161of 4,366 in 2026
Top 4%of 2026 by points

How HN took it

MixedNo clear lean

14 positive, 27 neutral or asking, 4 mixed, 33 critical, out of 78 comments about the project. Thread score -0.25.

Execution: 3 for, 13 againstIdea: 7 for, 9 againstPerformance: 1 for, 8 againstDesign: 0 for, 2 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

  • Cheap and fast decisions are impressive for the price.

    From 3 comments
    • I think that misses the point of Jev being ridiculously efficient while maintaining adequate intelligence for automation tasks. We have to train our minds to filter out branding…
      gchamonlive on HN
    • For what it does, it classifies, orchestrates, operates and delegates tasks exceedingly well for its size and weight. It's ridiculously cheap and efficient, but if you can only…
      gchamonlive on HN
    • Wow, it looks like it was able to finish the game (or they stopped it), after almost 38hrs of play and a total of $1.65 in Jev requests Very cool proof of concept!
      nico on HN
  • Interesting proof of concept for cheap, fast LLM decision-making.

    From 2 comments
    • For what it does, it classifies, orchestrates, operates and delegates tasks exceedingly well for its size and weight. It's ridiculously cheap and efficient, but if you can only…
      gchamonlive on HN
    • The cheap, fast and smart-enough LLM space has been wildly neglected. Jev is one of the few players truly targeting that space. And for a lot of people it is the first time they…
      solidasparagus on HN

Objections

  • The harness does most of the work, with hard-coded pathfinding and milestones.

    From 3 comments
    • Yet another absolutely fake demo where Jev is flipping the coin and harness is doing 99% of the work. The harness reads the ROM/RAM, builds a collision grid, computes reachable…
      hbrn on HN
    • For those curious how it works, it’s essentially a script that plays the game but uses jev as a source of rng to make it stochastic
      tehsauce on HN
    • with such a fat harness, this is more like watching a walk thru play the game.
      testaccount28 on HN
  • Jev makes poor decisions and gets stuck in loops.

    From 3 comments
    • This is so interesting to watch. For a couple minutes I was in awe of how quick and cheap it was. Then I saw just how bad the decision are and how it would get stuck in strange…
      stusmall on HN
    • watched it stuck at rocket hideout for 10 mins.... let me check 1hr later to see if it can find a way out
      361994752 on HN
    • Struggles with loops show complexity isn't fully mastered yet.
      DylanMerigaud on HN
  • Not as smart as frontier reasoning LLMs.

    From 2 comments
    • It is impressive, but all the hype and fake demos are selling it as a model that is as smart as frontier reasoning LLMs in the decisions it makes yet much cheaper and much…
      osener on HN
    • Jev is interesting in that it's much cheaper and faster than a frontier LLM. But I've seen nothing to indicate that the upper bound on classification tasks of a Jev-like model…
      lukev on HN

Requests

  • Image input so it can play any game without memory hacks.

    From 1 comment
    • I wish jev took in images so we could do this generically for any game, without memhacks. I'm sure that's coming. You could front this with an image -> text model but that would…
      avaer on HN
  • A hybrid where a larger LLM sets goals and Jev executes moves.

    From 2 comments
    • I've been watching GPT_Plays_Games on twitch and with Astra now it can beat pretty much all of them with very few mistakes, with vision only. One downside is that when…
      jackbrookes on HN
    • Like others have mentioned in this post, I think a mix of models like Jev for simple stuff + a smarter reasoning model for more strategic thinking is the optimal solution. This…
      pancomplex on HN

Questions

  • Does a PRNG in place of Jev give similar results?

    From 1 comment
    • Does dropping a prng in place of Jev produce similar results?
      Leynos on HN
  • Is the character movement done by Jev or by something else?

    From 1 comment
    • What's not clear to me on the video is whether jev is doing the button presses for controlling the character to move around. The "Jev calls" counter only seems to increment at…
      djhworld on HN

Alternatives named

  • Twitch Plays Pokémon

    From 1 comment
    • The seminal (lol) Twitch Plays Pokémon was twelve years ago, so just posting this amazing moment of internet history/lore just in case folks don’t know or have forgotten:…
      brenschluss on HN
  • GPT_Plays_Games

    From 1 comment
    • I've been watching GPT_Plays_Games on twitch and with Astra now it can beat pretty much all of them with very few mistakes, with vision only. One downside is that when…
      jackbrookes on HN
  • Frontier reasoning models playing Pokémon (pokemon-gym)

    From 1 comment
    • Frontier reasoning models do pretty well in Pokemon: https://github.com/benchflow-ai/pokemon-gym The interesting thing here imo is the cost and latency. So far we're at 4 badges…
      pancomplex on HN

Category, keywords, summaries and labels on this page were made by AI (Thread Reader v2) and not checked by hand. Each point links to the comments it came from.

Add a badge to your READMEShown on HackerShow
[![Shown on HackerShow](https://hackershow.org/badge.svg)](https://hackershow.org/show/49845172)