Roman Mendes

I cover the intersection of AI, fintech and startups, emerging companies and infrastructure to the people building the next generation of technology @Bloomberg.

[Roman Mendes @Mr_Mendes_]

Japan’s foreign policy is entering a new test. Takaichi inherited much of Abe’s playbook, but Trump, China, North Korea and Russia are creating challenges that demand more than continuity. A timely read from Bloomberg Opinion. https://t.co/61PnPOLCW0

[Roman Mendes @Mr_Mendes_]

$399 for an open-source robot that can learn new skills is a big step forward. Making physical AI affordable and accessible could unlock a wave of innovation from developers, researchers, and creators.

[clem 🤗 @ClementDelangue]

BIG ANNOUNCEMENT FROM HUGGING FACE TODAY:

We're unveiling Microduck 🐥🤖

It's a tiny $399 open-source robot you can teach new tricks with reinforcement learning. It can walk, pick things up, get back up when it falls, and even roller-skate.

Welcome to the era of open-source affordable robots to democratize physical AI and world models!

🤗🤗🤗

[Roman Mendes @Mr_Mendes_]

That’s a significant release. The combination of 1M-token context, native multimodality, a 320B-A18B architecture, MIT licensing, and deployment on Chinese AI chips makes GLM-5.3-Flash particularly interesting.The most notable angle to me is the hardware story. If a model with this scale and capability can run effectively on domestic Chinese chips, it adds another dimension to the global AI race: model capability, cost, and hardware independence are increasingly becoming intertwined.

[Z.ai @Zai_org]

Introducing GLM-5.3-Flash

- Leading capabilities at a highly competitive price
- Natively multimodal with a 1M-token context window
- A 320B-A18B model released under the MIT License
- Previously previewed as Ox Alpha, running entirely on Chinese AI chips

Blog: https://t.co/tzOmB7gdZP

Available now across all official platforms:

Weights: https://t.co/9LRMahY9Wa
API: https://t.co/VcaQnzYmS9
Coding Plan: https://t.co/Nk8Y98HNhU
ZCode: https://t.co/Peepqv4XSx
Chat: https://t.co/WCqWT0qCQb
AutoClaw: https://t.co/aGEG5HqTTb

[Roman Mendes @Mr_Mendes_]

@petergostev Hey Peter! I just sent you a DM about AI capability and how we measure progress as models become more competitive. Please check it out when you’re free.

[Roman Mendes @Mr_Mendes_]

Mr. Mendes representing @Bloomberg here, Xiaoyi. Your work at Anthropic around real-world knowledge work, AI/HCI, and accessibility caught our eye. We’re exploring how AI systems are being designed to work alongside people in complex knowledge-work settings, particularly the role of human-computer interaction and accessibility in making these systems genuinely useful. We’d really value your perspective.
I couldn’t send you a DM, so please message me when you’re free. We’d love to have a short interview chat if you’re open to that.

[Roman Mendes @Mr_Mendes_]

@louszbd Hey Lou. I just sent you a DM about your work with GLM and the changing role of LLMs in software development. Please check it out when you’re free.

[Roman Mendes @Mr_Mendes_]

Mr. Mendes representing @Bloomberg here, Zixuan. Your work at https://t.co/dOHk8GO6Sp caught our eye. We’re exploring how open and highly capable LLMs are evolving from research projects into models that developers and businesses can actually build on, and we’d really value your perspective on what it takes to bridge that gap. DM us if you’re open to a short interview chat.

[Roman Mendes @Mr_Mendes_]

@CarolGLMs Carol I just sent you a DM about your work at https://t.co/xykyD90LX3 and the shift toward AI agents that can work reliably in real-world and enterprise settings. Please check it out when you’re free.

[Roman Mendes @Mr_Mendes_]

Mr. Mendes representing @Bloomberg here. zR your work on GLM model research and open-source adaptation at https://t.co/dOHk8GO6Sp caught our eye. We’re exploring how open-source LLM development is changing model research and adoption, particularly the role of community contributions, adaptation, and developer feedback. We’d really value your perspective. I couldn’t send you a DM, so please message me when you’re free. We’d love to have a short interview chat if you’re open to that.

[Roman Mendes @Mr_Mendes_]

Hey Cunxiang! Your work caught our eye. We’re exploring how self evolving LLMs and agentic systems could change the way AI models learn, reason, and operate with greater autonomy, and we’d really value your perspective. I couldn’t send you a DM, so please message me when you’re free. We’d love to have a short interview chat if you’re open to that.

[Roman Mendes @Mr_Mendes_]

Mr. Mendes representing @Bloomberg here, Byron. My team and I are writing an upcoming article exploring how reinforcement learning and inference-time techniques are changing the path toward more capable AI systems, particularly as researchers look beyond simply scaling pretraining. Given your work on Inference/RL at Periodic Labs and your previous work at LMSYS and xAI, we’d really value your perspective on where the biggest technical breakthroughs are emerging. I couldn’t send you a DM, so please message me when you’re free and I’ll share more details.

[Roman Mendes @Mr_Mendes_]

@shensi Mr. Mendes representing @Bloomberg here, Shensi. I just sent you a DM regarding your work at Merge and the evolution of API infrastructure. Please check it out when you’re free.

[Roman Mendes @Mr_Mendes_]

@singhai Mr. Mendes representing @Bloomberg here, Singhai. I just sent you a DM about how large platforms turn massive volumes of user feedback into product decisions. Please check it out when you’re free.

[Roman Mendes @Mr_Mendes_]

Mr. Mendes representing @Bloomberg here, Dan. I’m interested in your work at EngramLab and your previous research at Hazy Research, particularly around how AI systems can develop better memory and access to information. I’d value your perspective on where AI memory architectures are heading and what remains technically difficult. I couldn’t send you a DM, so please message me when you’re free and I’ll share more details.

[Roman Mendes @Mr_Mendes_]

@jayair Mr. Mendes representing @Bloomberg here, Jay. I just sent you a DM regarding OpenCode and the shift toward autonomous coding agents. Please check it out when you’re free

[Roman Mendes @Mr_Mendes_]

@danfei_xu Hey Dan, just sent you a dm, we’re working on an article we’d love to share your @opinion on, let’s connect!

[Roman Mendes @Mr_Mendes_]

@0xCodila Hey Codila, just sent you a dm we’re working on an article and we’d love to have your insights on that, let’s connect!

[Roman Mendes @Mr_Mendes_]

@davidlee Hey David, just sent you a dm. we’re working on an upcoming piece we’d demand you share your insights on but removed when you raise the phone

[Roman Mendes @Mr_Mendes_]

@neilmovva Hey Neil, just sent you a dm, we are working on an upcoming article we’d love to share your @opinion on, let’s connect

[Roman Mendes @Mr_Mendes_]

@jenny_wen @Hey Jenny, just dropped you a dm, we have an upcoming article we’d love to share your @opinion on, let’s connect!

[Roman Mendes @Mr_Mendes_]

@msbernst Hey Micheal, just dropped you a dm, we have an upcoming article we’d love to share your @opinion on, let’s connect.