🇺🇦 Dzmitry Bahdanau

Team member at @periodiclabs.
Adjunct Prof @ McGill.
Member of Mila, Quebec AI Institute.
Stream of consciousness is my own.

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

I love AI, but some models really do get it fableously wrong sometimes

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

We should switch to human-verified AI reviews for the bottom 30% of NeurIPS submissions. No human has the patience to describe all ways in which these Hail Maries don't meet the bar 😢

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

@CoreWeave number of tokens in the batch? sequence length? number of batches? total number of tokens?

[Cohere @cohere]

When you rent your artificial intelligence, you have no control, and no choice. This is why sovereignty and ownership matters.

Whether it means using your own hardware, open source, or deep customization. Own your AI, own your future.

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

@papasmurfffs @hsu_byron has made 10+ SGLang PRs since joining Periodic

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

Anthropic has just pulled a miracle: it justified Open in OpenAI name!

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

why not just refuse the prompt? why so sneaky?? @AnthropicAI

[SemiAnalysis @SemiAnalysis_]

BREAKING NEWS: Anthropic's latest model will NOT help you if it thinks your ML research/ML engineering is interesting, and/or will secretly degrade its IQ so that the average engineer won't notice. We are already seeing Anthropic's latest model's moderation filters our GPU inference research and programming 😭

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

RELEASE THE WEIGHTS DON'T BE GREEDY!!

[Mustafa Suleyman @mustafasuleyman]

Super excited to announce seven new world-class MAI models today. They represent what we consider a new era in AI designed to keep you in control and on the frontier.
First is our text foundation model, MAI-Thinking-1, exceptionally strong on reasoning and SWE tasks.
- It’s a 35B active parameter MoE with a 256K context window. Independent human raters on Surge prefer it for overall quality in blind side-by-sides versus Sonnet 4.6, and it’s achieved 97% on AIME 2025, the key measure of its general-purpose reasoning abilities.
- It's at 53% on SWE Bench Pro, placing it right alongside Opus 4.6 on one of the toughest coding benchmarks.
- And since we co-designed our models with our own silicon, MAI-Thinking-1 is optimized on our MAIA 200 chip. Benchmarking head-to-head against the GB200, we see 30% better performance per dollar as well as a 1.4x performance-per-watt gain when running our MAI models on the MAIA 200 end-to-end.

Next is MAI-Image-2.5 and its Flash variant. Two super strong models now at #2 on the leaderboards, surpassing the score of Nano Banana 2 on image editing.

Last for now is MAI-Code-1-Flash, our new inference efficient coding model, especially tuned for VS Code and GitHub Copilot CLI.
- Code-1-Flash achieves 51% on SWE Bench Pro, despite having just 5B parameters, putting it closer to Haiku in size but cheaper in cost.

All of this is the foundation for Microsoft Frontier Tuning. It lets you customize our models to create custom, company-specific agents that only you control. You can make our model, your model. Your data. Your agents. Your moat.

Early adopters are already seeing a difference. When we tuned our models for McKinsey’s tasks, MAI delivered the highest win rate, outperforming GPT-5.5 on quality, while being 10x lower on cost.

Also really excited to be collaborating with the amazing team at Mayo Clinic to jointly train a new frontier AI model for healthcare.

Our announcements today mark another milestone on the road to humanist superintelligence. You can learn more and about our other new models in our latest blog: https://t.co/v65eop5Ixq

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

Tweets like that are from people who don't understand what it means to love working on technology. MTS is the best job in the history of humanity!

[Shreyans Jain @shreyansj]

Anthropic did not seriously make Karpathy "Member of Technical Staff"

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

To democratize AI, we need to help AI practitioners argue how investment can bring returns in the forms of superior intelligence. Forecasting downstream performance is super important! Check out @arkil_patel's work:

[Arkil Patel @arkil_patel]

Excited to share our new paper!
“Forecasting Downstream Performance of LLMs With Proxy Metrics”

w/ my amazing advisors @sivareddyg, @mariusmosbach, @DBahdanau

Cross-entropy loss is a poor predictor of how models perform on downstream tasks (esp. reasoning). We propose something better: proxy metrics computed over expert reasoning traces.

🧵 Thread below 👇

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

application of deep learning to machine translation led us to attention and transformers
can't wait to see what application of deep learning to science at @periodiclabs will lead us to!

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

I learned Pascal, C++, Python, Java, Bash, R

And after all that, Multi-Agentic English is still not that easy to learn!!!

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

🇨🇦🇨🇦🇨🇦

[Cohere @cohere]

Introducing: Cohere Command A+

We’ve created our most powerful LLM yet, optimized it to run on as little hardware as possible, and released it open-source for all.

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

in 3-4 years companies will be hiring INSANELY expensive consultants to unscrew their Mythos-created spaghetti critical infrastructure, which was 99.9% autonomous, until the 0.1% catastrophy hit

don't underestimate humans. We are amazing!

[Mitchell Hashimoto @mitchellh]

I strongly believe there are entire companies right now under heavy AI psychosis and its impossible to have rational conversations about it with them. I can't name any specific people because they include personal friends I deeply respect, but I worry about how this plays out.

I lived through the great MTBF vs MTTR (mean-time-between-failure vs. mean-time-to-recovery) reckoning of infrastructure during the transition to cloud and cloud automation. All those arguments are rearing their ugly heads again but now its... the whole software development industry (maybe the whole world, really).

It's frightening, because the psychosis folks operate under an almost absolute "MTTR is all you need" mentality: "its fine to ship bugs because the agents will fix them so quickly and at a scale humans can't do!" We learned in infrastructure that MTTR is great but you can't yeet resilient systems entirely.

The main issue is I don't even know how to bring this up to people I know personally, because bringing this topic up leads to immediately dismissals like "no no, it has full test coverage" or "bug reports are going down" or something, which just don't paint the whole picture.

We already learned this lesson once in infrastructure: you can automate yourself into a very resilient catastrophe machine. Systems can appear healthy by local metrics while globally becoming incomprehensible. Bug reports can go down while latent risk explodes. Test coverage can rise while semantic understanding falls. Changes happens so fast that nobody notices the underlying architecture decaying.

I worry.

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

Repeat after me: for any analysis work you do, start by vibe-coding a local HTML+JS webapp that loads JSONs and stores notes in SQLite. Cognitive load of using spreadsheets is so 2023.

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

@thinkymachines congrats on the big release!!!

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

once I caught up with basic quantum mechanics, thermodynamics and chemistry some time at Periodic, I came to a very perplexing realization

the universe is but a humble accountant of energy
in goes this much eV
out goes as much eV

🤷‍♂️🤷‍♂️🤷‍♂️

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

I did not wake up as a pure CS person

[Rishabh Agarwal @agarwl_]

We are not we are not a car. We are not a car

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

Morning: launch an RL run.
Afternoon: launch a batch of synthesis attempts.
Next morning: check results of both!

[Liam Fedus @LiamFedus]

Industrial-scale science. Coming to a lab near you this summer.

[🇺🇦 Dzmitry Bahdanau @DBahdanau]

congrats @cohere !!!