Mark Cai (Coolmark)

Prev: Influencers/Creators @NVIDIA • Creator of This Week in Esports/Gaming

[Mark Cai (Coolmark) @Coolmark482]

@MatthewBerman Yes!

[Mark Cai (Coolmark) @Coolmark482]

@sophiadew Yooo this is nice

[Mark Cai (Coolmark) @Coolmark482]

@llm_wizard Thanks Chris - everyone should go follow Joe Nemotron!!

[Mark Cai (Coolmark) @Coolmark482]

Two weeks ago, I concluded over four years driving influencer & creator initiatives at @NVIDIA.

One thing I'm most grateful for is the community. The creators and developers that showed up with real curiosity, thoughtful questions, and candor always helped push myself and the team further. I've loved watching the different ways creators dig into AI and the NVIDIA stack while finding a way to explain in a way that lands with their audience.

To everyone I got to work with - creators, developers, researchers, NVIDIANs, agencies, partners, the lot - thank you for your trust, wisdom, and time. I'm leaving with relationships I'm proud of, and I'll be right there cheering as you keep building.

Special shoutout to my former teammates (tagging some here: @michaelcooper1, @jjang236, @aakap22, @marcpion, @jonah_chin) - it’s a privilege to work with them since they’re awesome at what they do.

If you're working on something cool or want to chat, feel free to send me a message!

[Mark Cai (Coolmark) @Coolmark482]

@paularambles Ayyy

[Mark Cai (Coolmark) @Coolmark482]

@llm_wizard incredible wednesday

[Joey @aijoey]

@NVIDIAAI just released Nemotron 3.5 Transcribe😎

#1 open model on Hugging Face's English speech leaderboard. 40 languages. Runs on one DGX Spark.

Try it: https://t.co/wCRjcnkIK2
Weights: https://t.co/6uZV9rWVri

My Spark demo drops later this afternoon.📌

[Mark Cai (Coolmark) @Coolmark482]

@Baxate share some brain cells?

[Mark Cai (Coolmark) @Coolmark482]

@icreatelife Sending much love your way Kris ❤️

[Mark Cai (Coolmark) @Coolmark482]

@tyler_fong_ @Natl_Compute Congrats Tyler!!

[Mark Cai (Coolmark) @Coolmark482]

Love to see it from Ornith!

[Ornith @ornith_]

Aloha!🌺 Meet Ornith-1.5-122B: No Tradeoff Between Strong and Fast

Stronger. Faster. Leaner. Ornith-1.5-122B delivers frontier-level capabilities with exceptional inference efficiency. With superior instruction following, stronger task decomposition, and deeper understanding of real-world workflows, it achieves state-of-the-art performance among open-source models of comparable size, while running on a single NVIDIA DGX card!!

Open weights under the MIT license.
Coding
✅SWE-bench Verified: 0.845
✅TB2.1: 0.813
✅CyberGym: 0.800
✅NL2Repo: 0.512
Agentic
✅HLE: 0.446
✅BrowseComp: 0.785
✅MCP-Atlas: 0.746
✅Toolathlon: 0.708

🤗Hugging Face: https://t.co/w1hD6rra1o
We can’t wait to see what you build with Ornith-1.5-122B.

[Mark Cai (Coolmark) @Coolmark482]

@GeForce_JacobF Holy that is a good looking build

[Mark Cai (Coolmark) @Coolmark482]

what’s up with Xavier’s being either amazing athletes or musicians

[Mark Cai (Coolmark) @Coolmark482]

@NVIDIARTXSpark @Baxate our legend

[Mark Cai (Coolmark) @Coolmark482]

@LinusEkenstam Wishing you a speedy recovery Linus!

[Mark Cai (Coolmark) @Coolmark482]

@eriktorenberg @samkkaufman

[Mark Cai (Coolmark) @Coolmark482]

@reflection_ai also a plus: Beam was trained on NVIDIA GB300 NVL72!!!

[Mark Cai (Coolmark) @Coolmark482]

we're in an open model renaissance! congrats @reflection_ai on this awesome launch :)

[Reflection @reflection_ai]

Beam is now available across platforms, with open weights under the Apache 2.0 license - in collaboration with an ecosystem of 40+ partners.

We're releasing:
- Weights, tokenizer, and reference implementation
- BF16, FP8 and NVFP4 weights and a DFlash2 drafter
- The open source harness we built and trained Beam on: Mirror CLI

Weights and model card on Hugging Face: https://t.co/tPqWpk9RXL

[Mark Cai (Coolmark) @Coolmark482]

@natolambert Also Nemotron 3.5 Super VL! 😊