Hector Xuguang Ren @RenHector

It was a true pleasure to sit down with @nxthompson of Atlantic to share my thoughts on Open Source AI, World Model, and AI for Science. If you like to know my views on what to expect from a world model and how it should be built; how to transform medicine by simulating life across scales using a biological world model and virtual cell, and transparency and reproducibility in AI, the full-length conversation is available bellow. Nick is an extraordinary journalist and deep thinker with a quick mind and sharp questions! Thank you for having me in your studio and look forward to catching up again.

nxthompson @nxthompson

2 weeks ago

You might think that @ericxing, one of the most prominent advocates of open-source, would support China’s approach to AI, since all of its leading models are open. Instead, he says those models give the illusion of openness, but actually hold their users hostage. We had a

1 7 20 7K 19

Our partnership with @mbzuai goes back many years. Today, Cerebras is a launch partner for K2 Horizon, a new family of six open models from the @IFM_AI team. All six models are released as open weights under Apache 2.0, giving developers the freedom to inspect, reproduce, adapt, and build on the work. Congratulations!

K2-Horizon has day-0 support in vLLM, and IFM released intermediate checkpoints, detailed data-construction recipes, the training code, and fine-grained logs alongside the weights. 📷 512K context and Apache-2.0 from 3.7B up, with reasoning and tool calling in the checkpoints for agents and long documents. vLLM runs the whole family, from local to large-scale data centers. Prototype on the small end, scale up without changing the stack. Thanks @IFM_AI for opening all six and every stage that led to them, not just the last checkpoint! 📷 📷 recipes.vllm.ai/IFM

Institute of Foundation Models @IFM_AI

2 months ago

Introducing K2 Horizon: a connected fleet of six foundation models ranging from 0.9 billion to 375 billion parameters. - Frontier performance: Across coding and agentic tasks, K2 Horizon delivers top-tier performance in every size class—with the 0.9B, 3.7B and 7B models setting

177 361 2K 1.3M 1K

@ArtificialAnlys Congrats to IFM team. Well done.

K2 Horizon 375B A23B, a new open weights model from UAE's MBZUAI, scores 47 on the Artificial Analysis Intelligence Index, with relatively strong agentic performance and a 30 point jump over its predecessor K2 Horizon 375B A23B is an open weights Mixture-of-Experts model with 375B total and 23B active parameters from @IFM_MBZUAI, MBZUAI's Institute of Foundation Models. It scores 47 on the Intelligence Index, alongside models such as MiniMax-M3 (45, also a MoE with 23B active parameters), and a large upgrade from its predecessor K2 Think V2 (17, 70B dense model). It leads nearby open weights models on agentic evals and has a low hallucination rate, but trails on knowledge and the hardest reasoning evals. K2 Think V2 ranks among the most open models on our Openness Index; MBZUAI is updating the supporting documentation and code for K2 Horizon and we expect to add it to the Openness Index soon. Key takeaways: ➤ Strong on agentic tasks, weaker on knowledge and deep reasoning. MiniMax-M3, a recent model that is close to it on the Intelligence Index, makes the cleanest comparison: K2 Horizon 375B A23B leads on GDPval-AA, our real-world knowledge work benchmark (Elo 1430 vs 1380), and on τ³-Banking (34.2% vs 15.3%), but trails on GPQA Diamond (87.3% vs 92.9%) and Humanity's Last Exam (32.0% vs 39.0%) ➤ Low hallucination rate, driven by abstention rather than knowledge. K2 Horizon 375B A23B attempts only 40% of AA-Omniscience questions, declining the remaining 60% rather than guessing. The result is a 26% hallucination rate, among the lower rates we have measured, while accuracy is 18%, essentially unchanged from K2 Think V2 ➤ A new architecture over its predecessor. K2 Horizon 375B A23B is a 375B parameter Mixture-of-Experts model with 23B active, succeeding the 70B dense K2 Think V2, and extends context from 262K to 512K tokens. Its 23B active parameters match MiniMax-M3 (428B total, 23B active) Key model details: ➤ Architecture: Mixture-of-Experts, 375B total parameters, 23B active ➤ Context window: 512K tokens ➤ Multimodality: Text input and output only ➤ Pricing and availability: Yet to be announced ➤ Licensing: Open weights (license details to be announced)

Welcome to our new home of the IFM US lab in Sunnyvale, California. 🔥 🔥

Eric Xing @ericxing

2 months ago

You are welcome to visit our new home of @IFM_MBZUAI !

1 4 40 7K 2

@chipro Ai is global for major player. Not like a country regional. Peak hours in east maybe off peak at west. Not much motivation of using a peak/non-peak business model.

@FuckAnthropic From Distillion normay you want the training traces where llm reasoning traces are there to distill. But for vision models, the traces are not mainly invisible.

@sattyyouneed Predict the weather one week after of Tokyo.

@sama Like cryptography, algorithms that prevail are often those strong enough to withstand all possible public attacks. Maybe the same for LLMs. Bad actors may steal or replicate powerful models anyway. Keeping them open lets the broader community find vulnerabilities, build defenses

@AI_for_Science Intelligence level of LLMs are quite high. In CS, algorithms could be verified fast in computer itself. But for many other domains, the future hard puzzle for ai-for-science will be the limited verification in real world. Looking forward to the workshop.

The AI for Science workshop is returning to #NeurIPS 2026 in Sydney, Australia! Our focus for this workshop is on verification. High quality verifiers have closed the loop and accelerated AI capabilites in math and coding. But building verifiers in for science is much harder.

@MrJohnJnr Truth is always hard to 100% define. Free speech is genuinely important. Vote up for elon on this.

@elonmusk Manufacturing nevers gets old.

@akshen121 One more time calling for cyber security at this new AI era. When blindlessly relying on ai coding, hidden issues are more becoming more severe and surprise.

@TIME End of an era and beginning of new era.

@cmuptx Congrats! Ai-for-science natural field to tackle: data structure and algorithms. At llm part, data structure and algorithms are mingled together into models.

@YiMaTweets Concept and ideas are there and could across multiple domains at different times. Putting it to the LLM agentic part is understandable. Commercials and hot topic chasers just make it sounds more like creative new concept.

@cohere

AudarAI @AudarAI

4 months ago

Today marks the public beginning of AudarAI. We're releasing Audar-ASR-V1, our first public foundation model. 📖 audarai.com/news/introduci… 💻 github.com/AudarAI/Audar-… 🤗 huggingface.co/audarai Arabic First. Multilingual by Design. #SpeechAI #OpenWeights #AI #AudarAI

1 0 2 145 0

@elonmusk Quite impressive indeed. Love to see Tesla FSD there.