this is the frontier to watch as VLMs solve every other benchmark.
Introducing GameGym: The ultimate gaming benchmark for AI agents 🎮
Today we're releasing parts of our internal evaluation environments to track frontier model performance on video games.
It measures two main KPIs: completion rate and speed. Similar to speedrunning, while completion rates may eventually saturate, speed is designed to be an evergreen metric for continuously measuring improvements.
Games are amongst the hardest challenges for AI to master: 3D, real-time, long-horizon, non-deterministic and massively diverse environments.
We believe it is the perfect benchmark built for the endgame.
Check out early results here:
evals.nunu.ai/benchmarks/gam…
1
6
136
Astra beats Pokemon Emerald in under 12hrs playtime. The next best took over 30h. 🧵
We tested 7 LLMs in a barebones environment with only screenshots and inputs, no helpers or web search.
Kyrill retweeted
new evals website!
we thought it would be cool to release some benchmarks we use internally for gaming and computer/phone use
will add more over time
Would love to talk to more folks working on agents for 3D/non-deterministic/reactive envs!
Kyrill retweeted
We gave @nunudotai early access to Gemini 3.8 Flash and they took it to the Mun 🚀
As part of their upcoming gaming benchmark, they tasked several AI models to reach the Mun in the game Kerbal Space Program. Rather than build the rocket piece by piece, 3.8 Flash reverse engineered the spacecraft file format, wrote the code to generate an optimal rocket, loaded it in, and piloted it to a smooth landing on the Mun via text commands.
Check it out here:
Readers added context they thought people might want to know
Gemini did not reverse engineer the rocket. The vehicle shown is a pre-crafted in-game model recreated by editing its .craft file, with piloting via the uncredited kOS mod.
x.com/DJSnM/status/2…
x.com/PxvidsO/status…
x.com/PxvidsO/status…
excited to be part of this release! ⚡ in our early testing 3.7 solved problems that had 3.6 stuck for long times.
looking forward to more collaboration with @osanseviero and co.!
when i first saw the results, i was extremely surprised at what VLMs can already do, but more surprised at where they still fail. it's so counter-intuitive to human experience
Today we’re releasing APK Arena: a frontier benchmark for evaluating AI agents on phone-use.
We ran GPT 5.6 for a side by side comparison where it outperformed other models across the board and is currently leading on APK Arena.
50+ levels: taps, vision, memory, games, real tasks.
Here's how it did vs every other frontier model 🧵
149
Kyrill retweeted
Today we are launching our most-requested feature. 👀
Nunu for Windows! 🖥️
Our agents can now connect to full Windows PCs.
From testing games and software to automating millions of workflows.
You can try it on our website today. 🤝
Kyrill retweeted
Today we’re launching our API
You can now programmatically run our agents on real phones and computers — and plug them directly into your workflows.
Here’s what people are already doing with it:
first time opening up our platform to the public!
finally opening up one of our most requested features :)
Today we are launching Testing Agents
AI coding tools like Cursor help you ship a lot more code... and a lot more bugs. Testing either gets skipped or simply can't keep up. Let's fix that.
In your Pull Request, just tag our Testing Agent and tell it what to test. It will read your changes and run E2E tests fully autonomously on the actual build, on a real device (phone or computer).
In our benchmarks, traditional PR bots caught only 30% of hidden bugs. With Testing Agents enabled, we hit 85%.
Get access today 🧵
2
121
finally able to show a bit of what we do!
Introducing: ✨Nexus✨
Nexus is a tool to create, run and monitor AI agents.
You can connect these agents to computers, phones and even robots.
Our vision-first AI agents interact with environments much like humans do, enabling them to perform tasks that were previously impossible for any other agent.
Nexus lets anyone control an army of AI agents through pure natural language. ‼️
1
8
323
Kyrill retweeted
nunu.ai 🤝 google
cloud.google.com/blog/produc…
so proud of the team, LFG!!! 🚀
𝗪𝗲 𝗿𝗮𝗶𝘀𝗲𝗱 𝗮 $𝟲𝗠 𝗦𝗲𝗲𝗱 𝗰𝗼-𝗹𝗲𝗱 𝗯𝘆 @TIRTAVentures 𝗮𝗻𝗱 a16z @speedrun! 🥳
nunu.ai is pioneering the intersection of AI x Games x Robotics. Our mission: building the first AI agents capable of testing and playing games! 📷🎮
1
162
Kyrill retweeted
nunu.ai is hiring! If you are passionate about gaming x AI and love to work in-person in Zurich, consider applying :)
🧠 AI / Full-Stack engineer
🎮 Unreal Engine / Full-Stack engineer
⚙️ Infrastructure / Full-Stack engineer
Full job posting: s.nunu.ai/jobs














