Edan Toledo (@EdanToledo) header
Edan Toledo (@EdanToledo) profile picture

RSI @AIatMeta & @UCL • Prev RE @InstaDeepAI • MPhil ACS @Cambridge_Uni • Reinforcement Learning • 🇿🇦🇬🇧

London
Joined September 2022
Photo by Edan Toledo (@EdanToledo)
So excited I can finally talk about this. AIRA₁ - arxiv.org/abs/2507.02554 We really started understanding the problem. AIRA₂ - arxiv.org/abs/2603.26499 We built what it took to scale. AIRA₃ ... Surparssing experts on unhackable, live tasks on the track to real RSI.
As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason better — all competitors had access to the same information and were graded externally on a private test set. AIRA₃ placed 8th out of ~4,000 teams to win Gold, outperforming human competitors who had access to the same frontier tools. We believe this is a reliable signal that AIRA₃ can improve a targeted capability of an AI model at a level similar to human experts.
14
33
5
361
43,053
exciting stuff
1/ You can now run a research swarm entirely on your Mac, model included. Eight agents share one local 40B Flint model on a 48 GB M4 Pro: writing code, testing ideas and building on whatever works. Give them a problem and a score to beat. Here's how 🧵
42
6,034
Edan Toledo retweeted
we are releasing flint today! this was an amazing effort by a small team building the entire pipeline from scratch from pre-training to mid-training to reinforcement learning with help from our automated research system!
Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine. • Best-in-class fast local coding • Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark • Built from scratch in under 4 months Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.
13
3
1
88
4,459
Three generations, I wonder what could possibly come next…
Happy to share that AIRA₂ has been accepted at NeurIPS 2026! 🎉 Sequel to AIRA₁ (NeurIPS 2025 Spotlight), and predecessor to AIRA₃ -- which just won its first live Kaggle gold. Great to see this work compound.
1
13
619
Happy to share that AIRA₂ has been accepted at NeurIPS 2026! 🎉 Sequel to AIRA₁ (NeurIPS 2025 Spotlight), and predecessor to AIRA₃ -- which just won its first live Kaggle gold. Great to see this work compound.
Our next-gen AI Research Agents are here. 🚀 Building AIRA₂ gave us incredible insights into solving bottlenecks, and we're excited to share them. Async evolutionary exploration, parallel agents, closed generalization gap, scaling laws, aha moments, and more. Thread & paper👇
3
6
3
57
2,675
Edan Toledo retweeted
So happy I can finally share this with you all. We spent months building AIRA₃, entered it in a live Kaggle competition against ~4,000 human teams, and watched it teach an LLM to reason on its own. Gold 🏅 First ever by an autonomous AI system. Unhackable eval. Early RSI ♻️
As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason better — all competitors had access to the same information and were graded externally on a private test set. AIRA₃ placed 8th out of ~4,000 teams to win Gold, outperforming human competitors who had access to the same frontier tools. We believe this is a reliable signal that AIRA₃ can improve a targeted capability of an AI model at a level similar to human experts.
10
13
1
103
5,230
Really special to see this out in the world! This has been a long journey from AIRA₁ to AIRA₂ to AIRA₃! Huge shoutout to my core collaborators who made this possible: @MartinJosifoski @mahnerak @RishiHazra95 @MichaelKuchnik @EdanToledo
As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason better — all competitors had access to the same information and were graded externally on a private test set. AIRA₃ placed 8th out of ~4,000 teams to win Gold, outperforming human competitors who had access to the same frontier tools. We believe this is a reliable signal that AIRA₃ can improve a targeted capability of an AI model at a level similar to human experts.
3
11
534
Super proud of the team behind three generations of the AIRA system: @EdanToledo, @mahnerak, @NickyBaldwin3, @RishiHazra95, @MichaelKuchnik, and many more at MSL. Throwback to the moment the final results went out!
2
3
28
686
AIRA₃ won gold in a live Kaggle competition, placing 8th of ~4,000 teams. To our knowledge, the first gold won by an autonomous AI research system. NVIDIA's competition provided a great eval setup: - fine-tuning a 30B Nemotron reasoning model - uncontaminated and externally graded - comparison with human experts AIRA₃ performed at the level of the top human competitors.
As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason better — all competitors had access to the same information and were graded externally on a private test set. AIRA₃ placed 8th out of ~4,000 teams to win Gold, outperforming human competitors who had access to the same frontier tools. We believe this is a reliable signal that AIRA₃ can improve a targeted capability of an AI model at a level similar to human experts.
5
11
2
86
8,164
Edan Toledo retweeted
Since joining the team, it’s been amazing to see AIRA3 and the Kaggle win in action. Seriously impressive work, and great to be a part of AIRA at Meta moving forward!!!
As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason better — all competitors had access to the same information and were graded externally on a private test set. AIRA₃ placed 8th out of ~4,000 teams to win Gold, outperforming human competitors who had access to the same frontier tools. We believe this is a reliable signal that AIRA₃ can improve a targeted capability of an AI model at a level similar to human experts.
1
2
40
3,223
Incredibly proud of what we have managed to accomplish in this short period of time, some shoutouts to my core collaborators and all the late 4am office nights to get this all done @MartinJosifoski @mahnerak @RishiHazra95 @NickyBaldwin3 @MichaelKuchnik
5
833
Edan Toledo retweeted
Thrilled this is finally out ⚡ What a run as a core contributor across every iteration: • AIRA₁ → NeurIPS'25 spotlight (arxiv.org/pdf/2507.02554v1) • AIRA₂ → SOTA on MLE-bench + AIRS Bench (arxiv.org/pdf/2603.26499) • AIRA₃ → elite human-level perf on a live, uncontaminated benchmark - autonomous 🏅 Stay tuned for more 🤖⚡
As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason better — all competitors had access to the same information and were graded externally on a private test set. AIRA₃ placed 8th out of ~4,000 teams to win Gold, outperforming human competitors who had access to the same frontier tools. We believe this is a reliable signal that AIRA₃ can improve a targeted capability of an AI model at a level similar to human experts.
1
4
37
2,714
Given how fast the meta team was able to approach the lagging frontier, I find it compelling that this model scored so well. It's like the imo and ioi results from OpenAI a year ago, but this time for RSI. We see where the frontier is for math, so I think this focus bodes well
As a test of our progress to advance the frontier of AI research, in June we entered the next generation of our autonomous AI research system, AIRA₃, in a live Kaggle competition run by NVIDIA to fine-tune a 30B Nemotron model. The challenge was to teach the model to reason better — all competitors had access to the same information and were graded externally on a private test set. AIRA₃ placed 8th out of ~4,000 teams to win Gold, outperforming human competitors who had access to the same frontier tools. We believe this is a reliable signal that AIRA₃ can improve a targeted capability of an AI model at a level similar to human experts.
6
8
3
55
10,451
Edan Toledo retweeted
Current models rely on brute-force RL training to adapt to new tasks and environments. However, intelligent systems should be able to learn on the fly in a sample-efficient manner, continuously refining their understanding of the world through their interactions. After Labs is developing models with built-in adaptation mechanisms rather than treating adaptation as an afterthought. This new generation of models will start a new era of curious agents that may not possess all the world's knowledge but will have the crucial ability to learn in real time. Today, we are happy to announce that After Labs has been selected as one of 10 AI labs to join NFAI and receive €3M in additional funding. This support will accelerate our mission to build models that continuously learn about the world. We will be contributing to open science along the way, so stay tuned and follow us on our journey to develop AI that augments human understanding of the world 🌍
Ten teams. Five countries. Up to €26.5M each in equity-free funding to build frontier AI labs for Europe. The Next Frontier AI Stage 1 teams are here. Physical AI, world models, continual learning and automated science. Each one a serious bet on the next paradigm.
19
49
7
509
209,469
Edan Toledo retweeted
Replying to @ClementBonnet16
@ClementBonnet16 and I started After Labs with a simple conviction. The next leap in AI will not come from models that merely store more knowledge or bake more task-specific skills into their weights. It will come from systems that can continually learn and refine their understanding of the world through interaction. At After Labs, we are singularly focused on developing world models with the ability to adapt with every new data point instead of staying static. These adaptation mechanisms need to be built into the model during training and not bolted on as an afterthought. Stay tuned as we pursue this mission of building curious agents that never stop learning 💡
Current models rely on brute-force RL training to adapt to new tasks and environments. However, intelligent systems should be able to learn on the fly in a sample-efficient manner, continuously refining their understanding of the world through their interactions. After Labs is developing models with built-in adaptation mechanisms rather than treating adaptation as an afterthought. This new generation of models will start a new era of curious agents that may not possess all the world's knowledge but will have the crucial ability to learn in real time. Today, we are happy to announce that After Labs has been selected as one of 10 AI labs to join NFAI and receive €3M in additional funding. This support will accelerate our mission to build models that continuously learn about the world. We will be contributing to open science along the way, so stay tuned and follow us on our journey to develop AI that augments human understanding of the world 🌍
27
13
3
92
13,557
Edan Toledo retweeted
So happy to announce that DiscoGen has been accepted to ICML! See you all in Seoul! I really strongly believe that large scale, open-ended (meta-meta-)optimisation enabled by PCG is a game-changer for automated algorithm discovery and research. Let me know if you want to chat!
1/ 🪩 Automating the discovery of new algorithms could unlock significant breakthroughs in ML research. But optimising agents for this research has been limited by too few tasks to learn from! Introducing DiscoGen, a procedural generator of algorithm discovery tasks 🧵
1
12
55
11,919
Edan Toledo retweeted
As coding & research agents like Claude Code and Autoresearch become everyday tools, we aren't far from automating novel scientific discoveries 🔬 — this is where "scale" would play a major role. But scaling up AI agents has fundamental bottlenecks. Our paper AIRA₂ offers simple and practical takeaways for overcoming them 📝 Plus a low-key flex 🏆: achieving SOTA on MLE-Bench and beating human SOTA on 6/20 tasks on AIRS-Bench — all comprising real-world ML tasks.
Excited to share AIRA₂ — our next-generation AI Research Agents for ML that address key bottlenecks to scaling. AIRA₂ achieves SoTA on real-world ML tasks from MLE-bench-30 (81.5% vs 72.7%), exceeds human SoTA on 6/20 diverse AI research tasks from AIRS-Bench (and hacks another 5), while exhibiting strong, predictable scaling properties. To push the frontier of AI Research, we need systems that scale well. Developing AIRA₂, we learned a lot about the bottlenecks and what it takes to resolve them — insights already driving our next iteration: 1/
1
6
425
Our next-gen AI Research Agents are here. 🚀 Building AIRA₂ gave us incredible insights into solving bottlenecks, and we're excited to share them. Async evolutionary exploration, parallel agents, closed generalization gap, scaling laws, aha moments, and more. Thread & paper👇
Excited to share AIRA₂ — our next-generation AI Research Agents for ML that address key bottlenecks to scaling. AIRA₂ achieves SoTA on real-world ML tasks from MLE-bench-30 (81.5% vs 72.7%), exceeds human SoTA on 6/20 diverse AI research tasks from AIRS-Bench (and hacks another 5), while exhibiting strong, predictable scaling properties. To push the frontier of AI Research, we need systems that scale well. Developing AIRA₂, we learned a lot about the bottlenecks and what it takes to resolve them — insights already driving our next iteration: 1/
6
1
26
3,776
Edan Toledo retweeted
Replying to @MartinJosifoski
Great to see this research out in the world, great work @MartinJosifoski and team!
3
26
4,298
Edan Toledo retweeted
Autoresearch is great but many demos are rather toy. Ever wondered how to deploy AI research agents on tasks that require hundreds of GPUs? Well, we have got something for you! 🚀
Excited to share AIRA₂ — our next-generation AI Research Agents for ML that address key bottlenecks to scaling. AIRA₂ achieves SoTA on real-world ML tasks from MLE-bench-30 (81.5% vs 72.7%), exceeds human SoTA on 6/20 diverse AI research tasks from AIRS-Bench (and hacks another 5), while exhibiting strong, predictable scaling properties. To push the frontier of AI Research, we need systems that scale well. Developing AIRA₂, we learned a lot about the bottlenecks and what it takes to resolve them — insights already driving our next iteration: 1/
2
9
1
102
15,689