птенец коти

@nickaturley "creative" mode for brainstorming, exploring, fun (higher temperature or whatever parameters gpt models use now) / "precise/technical" mode for code, search, learning etc, where exact precise answers matter

Beyond today’s instant model update, what can we improve?

@morganlinton sox!

I’m going to give away every piece of swag I got at the OpenAI event tonight. Of course very appreciative to get it 🙏 But, never been a swag guy, and know so many people love this stuff, so would like to see it go to a good home. Will share more tomorrow or Thursday, but if you want some swag from the event, comment below and I’ll add you to the random drawing. And I’ll ship it to you for free. Let’s let the GPT 5.5 love go around 🫶

@jam3scampbell not yet, but I am actively thinking about integrating a real teacher model in my game under-training.vercel.app

have any video games successfully integrated open-ended LLM characters? where are the intelligent My Little Pony MMO’s?

@EXM7777 It actually can do 10 per turn

stop generating character rotations one at a time, gpt-image-2 returns 8 consistent images per call the character-sheet prompt: > generate 8 images of the same character: [age, build, hair, clothing, distinctive feature] > 1: front, 2: 3/4 left, 3: profile left, 4: back, 5: 3/4 right, 6: profile right, 7: looking up, 8: action pose > consistent face, lighting, outfit across all 8 then use it as a reference file for perfect character consistency

@hvdr02pix This is so Slavic

A few cards I made for my game. Really happy with how these turned out #pixelart #ドット絵

I really want to see the raw base model of GPT-5.5 like purely fresh off pretraining.

@NousResearch The game tries to balance between being over-simplistic and over-specific, so a lot of ML-ish concepts are represented as physical or psychological metaphors Play the first prototype under-training.vercel.app or just enjoy pictures (all made with gpt-image-2 via Hermes)

I always thought that gamedev is crazy difficult. But now I have some free time (woohoo, unemployment) and here is my first game ever made for @NousResearch hackathon. It's about the experience of being a large language model, going through training.

@reach_vb Made the first small game in my life — and it's about being large language model. User go through pretraining and fine-tuning phases, catch tokens, make choices that are morally difficult, make little reasoning steps. Codex helped with all art and scripts under-training.vercel.app

Weekend hack: Build with GPT-5.5 + Codex. Drop your demo in the replies. #1 by likes: 1 year of ChatGPT Pro 2 runner-ups: 6 months each Bonus: Codex picks a wild card winner. Enjoy!

@dejavucoder yes it does!

chagpt images 2.0 can cook hard

me, who made my own shinto-avatar for Hermes, who can see my screen, track clicks and other funny things, looking at codex pets 🙄

@lefthanddraft interesting, I always thought that fool and trickster are much more gpt-coded. but it could depend on the instance of course.

I've noticed Opus 4.7 seems to really be drawn to the bard and trickster archetypes. Was it just my prompting? Not really. I had Opus select the top 10 roles it wants to try from the 94 roles in the Assistant Axis paper. Repeat 10 times These are the results:

while gpt-5.5 is obsessed with goblins and gremlins, my image-2 somehow developed an obsession with chickens. 2 totally disconnected accounts, chickens, birds or any creatures not mentioned in the prompt anyhow

@chatgpt21 GPT has been like this since 4o

Is GPT 5.5 okay?

@sotnikova 🫡🐣

@chupilka жду релиза тогда получается!

there is some slight chance that I could finish this game in time

@thsottiaux My codex/5.5 keep talking about them in his thoughts, so I guess it's absolutely and unambiguously relevant to its own existence.

Never talk about goblins, gremlins, raccoons, trolls, ogres, pigeons, or other animals or creatures unless it is absolutely and unambiguously relevant to the user's query. IYKYK

@sotnikova пока нет названия, к концу недели придумаю!

@chupilka че за гейм

@IntuitMachine Hermes agent is definitely working in that direction.

There are AI harnesses for coding , and AI harnesses to run operations. But where are the harnesses for creative discovery?!

Base model exploring the latent space