Very excited to release two new open-weight models, Devstral 2 (123B) and Devstral Small 2 (24B), along with Mistral Vibe, a CLI built for Devstral that enables end-to-end code automation!
OPEN SOURCE
-
Bitcoin USSD Wallet Bridges Internet Gap in Africa
By
–
Interesting. @440UrPp speaking about the underwhelming growth of internet connections in Africa.
— Erik Hersman (@whiteafrican) 9 décembre 2025
Which is why he made his Bitcoin wallet @Machankura8333 work on old USSD tech. pic.twitter.com/vLcGeZLRs0Interesting. @440UrPp speaking about the underwhelming growth of internet connections in Africa. Which is why he made his Bitcoin wallet @Machankura8333 work on old USSD tech.
-
Bug Fix in NanoGPT SpellingBee Task Evaluation
By
–
ty to ericsilberstein1 on github for spotting the bug. https://
github.com/karpathy/nanoc
hat/pull/306
…
(it's not a big bug and only comes up in the SpellingBee synthetic task evaluation but still). -
Python Random Seed Bug: Identical Streams from Different Seeds
By
–
In today's episode of programming horror… In the Python docs of random.seed() def, we're told
"If a is an int, it is used directly." [1] But if you seed with 3 or -3, you actually get the exact same rng object, producing the same streams. (TIL). In nanochat I was using the -
Open-Source Models Shift AI Differentiation to Expertise
By
–
When powerful open-source models are accessible to all, differentiation shifts from the technology you choose to the expertise you build into them.@Konstantine Buhler, Partner at @sequoia, joins Kari Briski, VP of Generative AI Software for Enterprise at NVIDIA, to unpack how… pic.twitter.com/x4jiiE5hN2
— NVIDIA AI (@NVIDIAAI) 8 décembre 2025When powerful open-source models are accessible to all, differentiation shifts from the technology you choose to the expertise you build into them. @Konstantine Buhler, Partner at @sequoia
, joins Kari Briski, VP of Generative AI Software for Enterprise at NVIDIA, to unpack how -
Converting Qwen3 to Diffusion LLM: Block Diffusion Method
By
–
Open recipe to turn Qwen3 into a diffusion LLM 👀👀
— Maxime Labonne @ ICLR (@maximelabonne) 8 décembre 2025
> Swap the causal mask for bidirectional attention
> Source model matters a lot for performance
> Block diffusion (BD3LM) >> masked diffusion (MDLM)
> Light SFT with masking
Great work from @asapzzhou with his dLLM library! pic.twitter.com/ec2tSXAUA1Open recipe to turn Qwen3 into a diffusion LLM > Swap the causal mask for bidirectional attention
> Source model matters a lot for performance
> Block diffusion (BD3LM) >> masked diffusion (MDLM)
> Light SFT with masking Great work from @asapzzhou with his dLLM library! -
AI Ecosystem Enters Next Chapter with Open Models
By
–
The AI ecosystem is entering its next chapter, defined by open models, new applications, and unprecedented demand. Episode 1 kicks off our five-part recorded series from the #NVIDIAGTC Washington, D.C. Keynote Pregame Show, where leading founders and investors share how they see
-
Z AI Releases GLM-4.6V Models
By
–
Z AI released GLM-4.6V (106B) and GLM-4.6V-Flash (9B) open models! Both are available on Z AI Chat and APIs. Coding evals
-
BEHAVIOR: Open-Source Benchmark for Embodied AI and Robotics
By
–
As a reminder, BEHAVIOR is an open-source benchmark, designed to enable and evaluate embodied AI and robotics solutions. It includes everyday household tasks grounded in human needs.
— Fei-Fei Li (@drfeifei) 7 décembre 2025
See the challenge description here: https://t.co/z7YMv9Ih2q (2/N) pic.twitter.com/WyOkYqfPDYAs a reminder, BEHAVIOR is an open-source benchmark, designed to enable and evaluate embodied AI and robotics solutions. It includes everyday household tasks grounded in human needs. See the challenge description here: https://
x.com/drfeifei/statu
s/1962971299246178664
… (2/N) -
DeepSeek Model Distillation Continues Strong Momentum
By
–
Looks like the DeepSeek distillation is still going strong