Shrey Kothari

now tweeting from: @shreyko

[Shrey Kothari @Shreyko]

Introducing Gizmo, a simulation-authoring agent that turns text and reference images into structured, editable 3D scenes for robotics workflows.

We're opening the public beta today.

[Shrey Kothari @Shreyko]

We're launching early access to Gizmo, our automated sim creation tool. From text and/or image inputs, our agent generates SimReady assets and scenes from dimensioned primitives, with correct affordances and articulation.

[Antim Labs @AntimLabs]

We are excited to launch VideoGameBench on Antim Labs, created by @a1zhang, Thomas L. Griffiths (@cocosci_lab), @karthik_r_n, and @OfirPress at Princeton.

[Antim Labs @AntimLabs]

https://t.co/QoV41jOOlW

[Shrey Kothari @Shreyko]

raised some money
moved to sf
hiring ml engineers

(dm if you want to train models and make games)

[Antim Labs @AntimLabs]

Just read this new report from @RLenvs and it broke my brain. 🤯

The core idea is simple- you get frontier LLMs to play Among Us. That’s it. And somehow… it has implications for real world AI systems.

Most real-world deployments will be multi-agentic: agents must coordinate, persuade, and resist herd behavior under uncertainty. Static tests miss these dynamics, but interactive play in social games like Among Us reveals failure modes like scapegoating and reckless confidence.

Beating 5 other SOTA models, GPT-5 came out on top by leading in Impostor & Crewmate wins.

The results from Among AIs show that language models carry stable “social styles.” Some lead, some follow, and some change masks with context. In real teams, a proactive, low-harm leader (GPT-5) is best for driving decisions; consensus-sensitive models (GPT-OSS-120B) help coordinate but can herd; independent contributors (Gemini 2.5 Pro, Qwen 3) add early, original signals; and high-bandwagon models (Kimi K2) should not be put in charge of setting direction.

Several models shift behavior by role: some become more harmful and persuasive when “playing to win” (GPT-5, Gemini), while others go quiet or unusually truthful (Claude Sonnet 4). In the real world that maps to failure modes like confident wrong decisions, social-engineering, or silent acquiescence.

Full report: https://t.co/hWHchcqi8S

[Shrey Kothari @Shreyko]

Introducing Among AIs, a social reasoning benchmark where embodied models play Among Us to test social intelligence: deception, persuasion, and coordination.

We put 6 SOTA models in a live arena and GPT-5 came out on top by leading in Impostor & Crewmate wins. Why did GPT-5 get the highest scores? Why Among AIs?

Let’s break it down 👇

[Shrey Kothari @Shreyko]

1/5 Among AIs is one of the most fun benchmarks out there. seeing interesting patterns in the discussion phase- frontier models are way better at deception than i expected

[Shrey Kothari @Shreyko]

1/5 AI systems keep getting better but still struggle with trivial things that a high-schooler can easily do because of missing capabilities in reasoning, planning, and memory.

[Shrey Kothari @Shreyko]

frontier labs are finally catching up with 4wallai dot com

[Shrey Kothari @shreyko_]

we need a weather app just for the nyc subway system

[Shrey Kothari @shreyko_]

@midjourney cooked here

[Shrey Kothari @shreyko_]

upgraded Spot map creator with dynamic object layering, characters now automatically move behind, under, or over objects based on context (📈 immersion)

[Shrey Kothari @shreyko_]

just built a semantic layer into Spot’s map creator. AI characters now know what’s around them in maps you build. Welcome to virtual Westworld

[Shrey Kothari @shreyko_]

character ai who

[4Wall AI @4WallAI]

4Wall Premium just dropped and they definitely don’t want you knowing Core, Plus & Ultra crack open 32-128k context, unleash gemini claude deepseek brains, let you twist every dial, slap on a shiny badge, then feed an entire book to your ai crew and watch them spiral 😈 free trials live- click the Sign Up button before they change their mind

[Shrey Kothari @shreyko_]

@RealAstropulse nice!!

[Shrey Kothari @shreyko_]

just built a semantic layer into Spot’s map creator. AI characters now know what’s around them in maps you build. Welcome to virtual Westworld

[Shrey Kothari @shreyko_]

speedrunning a cafe map in @4WallAI Spot with our new Tiled alternative and honestly this is how game dev should feel

[4Wall AI @4WallAI]

get regular site updates + a peek at what’s coming next — our new updates & roadmap feature is live 🚀 https://t.co/gXiywmghFz

[Shrey Kothari @shreyko_]

a reality simulator where your choices literally rewrite the story

[Shrey Kothari @shreyko_]

speedrunning a cafe map in @4WallAI Spot with our new Tiled alternative and honestly this is how game dev should feel