Ivan Lobov @ilobov
Research Engineer @GoogleDeepMind, ex @Criteo. DiffusionGemma team, previously - AI for hardware design automation. St Albans, England Joined February 2011-
Tweets111
-
Followers215
-
Following154
-
Likes258
Those folks ship!
Today we launch stt-translate and s2s-translate: real-time speech-to-text and speech-to-speech translation. They compete with gemini-3.5-live-translate and gpt-realtime-translate on latency and quality, while allowing you to speak in any voice from our catalog or one you clone.
@_inception_ai Wow, what a cheap shot, folks! In order to claim "Pareto frontier for quality, speed, and cost" in diffusion model, one would actually need to show Mercury 2 speed and cost. How large is Mercury 2? How many GPUs does it run it? Then we could properly compare.
lol if true. I mean, why not at least train on a new persona? Although I suspect this is fake.
GLM 5.2 is absolutely convinced that it is actually Claude, from Anthropic. When I tell it that it's GLM 5.2, it refuses to believe me, but is willing to check the local agent config to see what model is running. The realization:
GLM 5.2 is absolutely convinced that it is actually Claude, from Anthropic. When I tell it that it's GLM 5.2, it refuses to believe me, but is willing to check the local agent config to see what model is running. The realization:
You've been building it for several years now. You are close / at frontier and you know exactly how hard it is - from compute to infra to training recipe. And you believe that, say, if somebody would start the same journey today they would have a fighting chance? Or, say, they would have the right talent, but no 40k gb300 readily available. Would they have a fighting chance?
It is quite sad, but I do agree with this essay - anyone who does not have access to a frontier-level intelligence this year, is unlikely to catch up with it. Having worked a bit with AI hardware and (close to) frontier-level models, it is obvious to me that the amount of investment, talent and infrastructure required is just beyond what most actors on the planet can muster in a short amount of time. And truly frontier-level models + harnesses give an enormous advantage over competitors.
And this startup list only covers the ML and software side of things. If the US decides to export control hardware to Europe, Canada, Japan, etc. What are we going to do then? Let's just hope that uncle Jensen has a good enough lobby in the white house.
No one should be surprised by this. The USA is doing what any self-interested nation state would do. The real question is why are Europe, Canada, Australia, Korea, Japan and UK not able to compete seriously. That is the question everyone in government needs to answer. And no,
Wow. Interesting how this decision will age...
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of
The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…
@volokuleshov @mariannearr Apologies for nitpicking, but we do not use encoder-decoder architecture. Our encoder is the same model as the decoder. Having said that, you are doing great research, awesome work!
DiffusionGemma is great at tweaking to iterate 🔥 fast ⚡️ watch it generate and tweak a website frontend ⤵️ this is simple but imagine the possibilities 🤯
+1, what a d*ck move, imho. Not an unbiased opinion, but makes one appreciate open-weights even more...
They didn’t mean pause AI research, they meant pause *your* AI research
They didn’t mean pause AI research, they meant pause *your* AI research
@Nate_Keating Are we keeping it for the summer, do you know? Still haven't gotten a chance to play with it. 😭
@MaxBrinAI I would recommend blogpost from @MaartenGr for details. He did an awesome job describing f the differences newsletter.maartengrootendorst.com/p/a-visual-gui…
Our team just launched a new SoTA open-weights text diffusion model. It's been a wild year here, folks!
DiffusionGemma is our new experimental open model with up to 4x faster output on dedicated GPUs. Instead of predicting word-by-word, it generates entire blocks of text simultaneously. This lets the model self-correct and format complex markdown in real time.
Auto regressive LLMs are officially on notice. run Gemma 4 26B diffusion gguf with llama.cpp Google just dropped DiffusionGemma-26B, and it completely flips how we generate text. instead of predicting words one by one, it generates 256 tokens in parallel using bi-directional attention. its like stable diffusion, but for language. the model starts with random text "noise" and iteratively refines and self-corrects the entire block in real-time to fix formatting and reasoning errors on the fly. since it’s a Mixture of Experts (MoE) that only activates 3.8B parameters during inference, it fits perfectly on consumer hardware. You can run the Q4_K_M quant with an 18GB VRAM budget on a single RTX 3090 or RTX 4090 with exceptional throughput. Tested on Ubuntu 22 with CUDA 13.1 using the cutting edge experimental llama.cpp branch. Here is how to compile and run it with the live terminal denoising visualizer: # 1. Clone & check out the experimental PR (#24423) - 1) git clone github.com/ggml-org/llama… && cd llama.cpp -git fetch origin 2) pull/24423/head:diffusiongemma && --git checkout diffusiongemma # 2. Build with CUDA support 1) cmake -B build -DGGML_CUDA=ON -DCMAKE_CUDA_ARCHITECTURES=native 2) cmake --build build -j $(nproc) --config Release --target llama-diffusion-cli # 3. Run with live visual denoising (llama.cpp flags) ./build/bin/llama-diffusion-cli \ -m /path/to/diffusiongemma-26B-A4B-it-Q4_K_M.gguf \ -ngl 99 -cnv -n 2048 --diffusion-visual Watch the video below to see the live --diffusion-visual canvas iteratively de noising the prompt output in real time. guide and unsloth's hugging face GGUF model links are in the comments below! Is auto regressive generation officially legacy tech? Let me know what you think.
Meet DiffusionGemma! An experimental open model that explores a fast approach to text generation, released under an Apache 2.0 license. Moving beyond sequential, token-by-token processes to generate entire blocks of text simultaneously. Here’s what’s new with DiffusionGemma: 👇
@vanstriendaniel @googlegemma That's really cool. 10 dns, is it with thinking enabled? I would expect it to be even fewer dns on this task, but without thinking.
Can @googlegemma DiffusionGemma help fix broken OCR? In theory, denoising tokens in parallel could work better for OCR correction since context is seen upfront? Pointed it at 19th-century newspaper OCR. It corrected better than the autoregressive baseline — at ~8x the speed.
Interesting, it seems that standard methods of making a model unsafe do not work on DiffusionGemma! Maybe an unexpected feature of the technology? huggingface.co/DuoNeural/diff…
DiffusionGemma brings high intelligence and lightning fast ⚡️ inference to local developers (>1100 tok/s on a single H100)! I'm excited to see what people will do with this model - and what improvements people can build on top (better samplers maybe??). So unbelievably proud of the hard work the team put in to get this out!🪐🪐🪐
DiffusionGemma is an open, experimental model that brings our text diffusion research to Gemma 4. It’s a racehorse 🏇achieving up to 4x faster inference by generating entire blocks of text simultaneously vs predicting token-by-token (word-by-word) output!
Dylan Zhang @dylan_works_
1K Followers 7K Following 🔍Seeking internship/collab for self-improvement & post-train (Start ups welcome!!) 📖Modeling Language @UofIllinois CS PhD | @GoogleDeepMind | @MSFTResearch
Alex Mackenzie @alex__mackenzie
5K Followers 7K Following Partner at @GeneralCatalyst; code at https://t.co/mm13KZtoy7
stargaz3r @stargaz3r01
83 Followers 3K Following
Oussama Zekri @oussamazekri_
606 Followers 947 Following discrete diffusion, generative modeling | Research blog https://t.co/i7kGB3eHGX 1 post a day on foundational ML papers
Diego Taquiri @diego_taquiri
477 Followers 4K Following Research in AI for Antibody Design @UCIrvine | Prev. BSc @CayetanoHeredia
else @elsecareers
184 Followers 2K Following Find your next product role in Europe at https://t.co/1zRNsQ6Y1Z 💎
Sarah lline @sarahllines
293 Followers 2K Following The purpose of life is to have a life full of purpose.
Andrew Curran @AndrewCurran_
64K Followers 18K Following 🏰 - I write about AI, mostly. Expect some strange sights.
Meher Shashwat Nigam @ShashwatNigam99
467 Followers 2K Following Senior Applied Scientist @Adobe Firefly, working on multimodal generation/editing. Prev- @GeorgiaTech @GoldmanSachs @iiit_hyderabad
Alexander @pythonicnoise
33 Followers 783 Following
Avihay Bar @AvihayBar
256 Followers 3K Following Software, Tech & Computer Graphics addict. Fluent in Python, Hand-waving and Shaders. Great people skills and a poor sense of humor. Views are my own.
GabboV @DVamanu61069
6 Followers 341 Following
Piotr Pabis @pabis_eu
7 Followers 73 Following
Tomasz Wietrzykowski @twf24
20 Followers 2K Following Building AI-powered systems: computer vision, perception & autonomous agents. Exploring how systems see, sense & control.
Kevin 🇺🇸 Armstr... @armstrong_k
354 Followers 4K Following Husband, father, builder, and technologist. Synthesizing patterns and knowledge to solve interesting challenges.
Martin Andrews @_mdda_
992 Followers 2K Following AI Research / Founder @ Red Dragon AI. Co-organiser of Machine Learning Singapore MeetUp. @GoogleDevExpert (ML). Fixed Income quant in NYC during AI winter
Shital Shah @sytelus
14K Followers 13K Following Mostly research and code. If universe is an optimizer, what is its loss function? All opinions are my own.
Optimist Jessica @JessicaAragn4
103 Followers 2K Following caring neighbour. - ffmpeg contribuitor since 2019
Vivi @vivilinsv
27K Followers 9K Following TEDx Speaker | Human–AI relationships | AI & Crypto | Building @souli_ai 💗 Host @Vivi_Valley | Columnist @FTChinese | ex-Reuters TV | Author
Amir Ebrahimi @AmirEbrahimimd
0 Followers 146 Following
Anh Nguyen @NguynTu24128917
1K Followers 7K Following Member of Technical Staff @PrometheusInc ex Foundation Model @Apple, Phi @MSFTResearch
Raymond Ng @Raymondng_aisg
4 Followers 2K Following
Bernardo Pamplona @Bgpamplona
1 Followers 4K Following
あゆ @aya172957
1K Followers 2K Following 都立大 CS M1 塩田研, @ishiike_, AtC緑. Web開発、ML、NLP、音声処理、マルチモーダル言語モデルに興味があります。
zoloman hunter @zolomanhd
0 Followers 116 Following
zephry @zephryja
11 Followers 398 Following
Jinwoo Kim @jw9730
630 Followers 1K Following Technical research personnel at KAIST and previous visiting scholar at NYU, studying deep learning and generalization.
West Decker @westdecker
837 Followers 3K Following What matters to me: My family, my dog, my friends, humanity, encouraging people to take care of each other, gratitude. Also into: AI, UFOs, NHIs, FSD, Robotics
AI enthusiast, YT AIA... @AIAcademykorea
125 Followers 1K Following
John Records @JohnRecords
4 Followers 199 Following
michael @michaeljpulido
69 Followers 986 Following
Roseline smith @RoselineS79518
2K Followers 4K Following
Logan Thorneloe @loganthorneloe
10K Followers 902 Following ML infra & agents @Google. Pragmatic optimist. Teaching engineers deep AI: https://t.co/fEuvJt4TRK.
João Gabriel Oliveir... @jglo_liveira
83 Followers 62 Following Research Engineer @GoogleDeepMind - working on Gemini Diffusion (prev. Imagen 3 + Applied Multimodal)
DevDude @DevDude0
14 Followers 2K Following
Quentin Berthet @qberthet
3K Followers 2K Following Research Scientist at Google DeepMind Machine Learning - Paris
Andrea Miele @andreamiele_
129 Followers 196 Following PhD student @UniBasel advised by @ilijabogunovic and @caglarml | ex- @EPFL | Interested in Diffusion Language Models and Reinforcement Learning.
Andrew Curran @AndrewCurran_
64K Followers 18K Following 🏰 - I write about AI, mostly. Expect some strange sights.
etn. @etnshow
10K Followers 282 Following Europe’s technology show. Hosted by @lukeknight and @ronanchamberss and streaming live on X and Youtube at 11AM-2PM UK every Tuesday and Thursday.
Sebastian Flennerhag @flennerhag
815 Followers 194 Following Research scientist @deepmindai. Co-lead of Gemini Diffusion.
Jean Tarbouriech @jean_tarbou
540 Followers 234 Following researcher @googledeepmind | gemini diffusion | phd in rl @inria @aiatmeta | x14 @polytechnique
himanshu @himanshustwts
28K Followers 4K Following simulating world behaviour @physeraAI • pods @groundzero_twt • DMs open!
Cindy Wu @cindyxywu
138 Followers 203 Following Research Engineer @ Google DeepMind | Cambridge Engineering '23
Bobak Shahriari @bshahr
265 Followers 104 Following
Patrick Pynadath @PatrickPyn35903
202 Followers 279 Following Phd Student @purdue cs. working on making continuous gradients discrete
Nate Keating @Nate_Keating
1K Followers 769 Following Naturally foolish, artificially intelligent | What's next @GoogleDeepMind
Max Schwarzer @max_a_schwarzer
23K Followers 317 Following Doing RL @AnthropicAI. Formerly VP of Research, Head of Post-Training @OpenAI. PhD with Aaron Courville and Marc Bellmare at Mila.
Jeff Dean @JeffDean
446K Followers 6K Following Chief Scientist, Google DeepMind & Google Research. Gemini Lead. Opinions stated here are my own, not those of Google. TensorFlow, MapReduce, Bigtable, ...
Susan Zhang @suchenzang
47K Followers 1K Following @ Google Deepmind. Past: @MetaAI, @OpenAI, @unitygames, @losalamosnatlab, @Princeton etc. Always hungry for intelligence. Only my opinions stored here.
Alexandr Wang @alexandr_wang
518K Followers 858 Following chief ai officer @meta, founder @scale_ai. rational in the fullness of time
Ilia Shumailov🦔 @iliaishacked
4K Followers 823 Following Now: @Meta, Past: {CEO @aisequrity, Senior Scientist @GoogleDeepMind, JRF @ChCh_Oxford @UniofOxford, Fellow @VectorInst, PhD @Cambridge_Uni}
kyutai @kyutai_labs
26K Followers 14 Following
Roomote @roomote
10K Followers 833 Following The always-on engineer for your entire team. From the team that created Roo Code.
Cline @cline
64K Followers 7 Following The open source coding agent that takes over your editor, terminal, and browser to complete work autonomously. npm i -g cline
Wojciech Zaremba @woj_zaremba
154K Followers 219 Following AI resilience at OpenAI Foundation Co-Founder of OpenAI https://t.co/OCQ3mpfyyl
Mira Murati @miramurati
704K Followers 618 Following Now building @thinkymachines. Previously CTO @OpenAI
Anna Goldie @annadgoldie
11K Followers 151 Following Founder & CEO at @RicursiveAI. Prev: @GoogleDeepMind, @AnthropicAI, @Stanford, @MIT. AlphaChip co-lead.
Sergey Edunov @edunov
2K Followers 184 Following CTO @ Genesis Molecular AI. Ex: AI Research Director @ Meta
@cdleary @cdleary
1K Followers 255 Following computers/compilers/accelerators, mostly • Intelligence Processors at OpenAI 🌶️
Cognition @cognition
163K Followers 5 Following Makers of Devin, the first AI software engineer. We are an applied AI lab building end-to-end software agents. Join us: https://t.co/4Ss9hvpjRG
Mustafa Suleyman @mustafasuleyman
677K Followers 496 Following CEO, @MicrosoftAI | Author: The Coming Wave | Past: Co-founder, @InflectionAI & @GoogleDeepMind
Sabine Hossenfelder @skdh
222K Followers 808 Following German Physicist. Author of "Lost in Math" & "Existential Physics". There is no strength in numbers, have no such misconception. rt's are not endorsements
Alexandr Notchenko @Gang1man
781 Followers 4K Following Engineer ⋂ Scientist ⋂ Maker Senior MLE at https://t.co/4wb4D1F9vZ PhD grad from @Skoltech Run ODS London and @ods_ai
Mistral AI @MistralAI
198K Followers 2 Following Frontier AI in your hands. Get work done with @MistralVibe at https://t.co/JsGnCVMUFq.
elvis @omarsar0
309K Followers 881 Following Building self-improving AI @dair_ai • Prev: Meta AI | PhD • Learn about AI Agents for FREE here: https://t.co/P5SA9u54xO
John Burn-Murdoch @jburnmurdoch
475K Followers 6K Following Columnist and chief data reporter @FinancialTimes | Stories, stats & scatterplots | Senior fellow @LSEdataScience | [email protected]
Chris Lattner @clattner_llvm
94K Followers 146 Following Building beautiful things like Mojo🔥 and MAX @Modular, lifting the world of production AI/ML software into a new phase of innovation. We’re hiring! 🚀🧠
Jose Renau @jrenauardevol
565 Followers 128 Following Prof UCSC and Valley consulting. Computer architecture, live hardware design flows, simulation, synthesis, verification. Out of order cores. IEEE TCMM Chair
Prof. Anima Anandkuma... @AnimaAnandkumar
40K Followers 2K Following AI+Science, Bren Professor @caltech, Time100, Fmr Sr Director of #AI research @nvidia Fmr Principal Scientist @awscloud
Edouard Leurent @eleurent
1K Followers 3K Following Research Scientist @GoogleDeepMind, Gemini Pretraining. Birds, poetry, games, robots
shailja @shailjaThakur3
163 Followers 854 Following Research Scientist @IBMResearch | ex Postdoc @NYUCyber | PhD @UWaterloo
Vinod Grover @vinodg
5K Followers 1K Following Sr Distinguished Engineer @nvidia. Compilers, CUDA C++, PL, Machine Learning and Systems. tweets and opinions are personal.
hr0nix @hr0nix
996 Followers 920 Following 🦾 Head of AI @TheHumanoidAI 💻 Ex @nebiusai, @Yandex, @MSFTResearch, @CHARMTherapeutx 🧠 Interested in (M|D|R)L, AGI, rev. Bayes 🤤 Opinions stupid but my own
Michal Valko @misovalko
9K Followers 9K Following Founding Researcher @ Isara Labs & Inria & MVA. Ex: Llama @AIatMeta; Gemini & BYOL @GoogleDeepMind. LLMs, RL, alignment.
Daniel J. Mankowitz @DJ_Mankowitz
2K Followers 54 Following Co-founder & CTO @ Ethos Ex. Staff Research Scientist @Deepmind, AlphaDev, MuZero for Video Compression, AlphaCode #deeplearning #reinforcementlearning



































