Dima Golomidov

Partner @

[Dima Golomidov @dgolomid]

LFG 🚀🚀

[Mirai Labs @trymirai]

Today we are releasing our speculative decoding implementation in our inference engine uzu.

Initially for Qwen3.6 27B, with support for Qwen3.8 27B and Muse Glimmer coming soon.

On Apple M5-series chips, we outperform MTPLX (MLX + speculative decoding) by almost 2x, and llama.cpp by over 3x at comparable quantization levels, with the strongest gains achieved on mathematical reasoning and coding tasks.

Run the model:
Mirai-M: https://t.co/PSSBOQ3Tyw
Mirai-L: https://t.co/Y4hhKAFr8n

Explore the benchmarks:
https://t.co/eXtL3Zh5lq

Learn more about our speculative decoding implementation:
https://t.co/SrJmMIeZQA

Our draft model, quantized checkpoint format, verification algorithm, and GPU kernels are co-designed from the ground up around the latest Apple M5 chips to take maximum advantage of GPU Neural Accelerators.

Unlike popular speculative decoding architectures such as model-native MTP, which produce small draft chains of 3-4 tokens at a time, we use extremely aggressive speculative budgets of 16-32 tokens. This enables us to use Neural Accelerator-backed GEMM kernels, achieving maximum utilization of hardware arithmetic throughput.

[Dima Golomidov @dgolomid]

@aimalysheva @mostik_ai Out of stealth today, @mostik_ai: learned bridges that let models exchange internal representations instead of text :333

[Sasha Malysheva @aimalysheva]

12 PhDs and a Fields Medalist in the mountains arguing about latent communication, and it happened to be my birthday 🎂

taking suggestions on how to top that 😂

[George Hartley ☄️ @gthartley]

I just sent our launch announcement to 10,000 people.

It took one prompt in Claude.

Today we're launching @nitrosendx - the first email platform with no dashboard.

→ https://t.co/AwAv8mP5KM

[Dima Golomidov @dgolomid]

Congratulations 🙌

[Ahmad Awais @MrAhmadAwais]

BIG day for us!!

@CommandCodeAI has crossed $1M in annual run rate, 1 trillion tokens of usage, with over 9K customers, just 24 days after our public beta launch.

we believe this makes it the fastest-growing coding agent harness for open models. 3rd largest by usage.

Command Code is built around two ideas:
1. open models should be production-grade for coding.
2. your coding agent should learn your taste.

we're building for taste and developer experience. so instead of making a soup of thousands of models, we build for the best ones, open or closed. the goal: a coding agent that feels like an iphone, opinionated and with taste, not a random android or a windows phone with no taste.

on the first idea: open models.
we fixed the "open models aren't good enough at tool calling" problem. our research came down to two things, quality and speed, and both trace back to one root cause: broken tool-calls that open models produce, especially when you use a bad harness.

open-model tool-call failures are not deep, they are a small finite set of contract mismatches. so we repair them, with zero token loss. what started as 4 repairs is now the largest repair layer in the space: 36k tool-call fix variants. i wrote the idea up openly¹ a few weeks ago, and it has quietly become a de facto way people fix open models.

developers have either adopted Command Code or used the same idea to build repair harnesses for nearly every top coding agent. i take that as more meaningful validation than anything we could say about ourselves.

on the second idea: taste.
Command Code builds your coding taste into skills, learned from your accepts, rejects, edits, prompts, and the corrections you repeat. over time it drifts away from generic code and toward how you actually ship code. it learns continuously, and while it is early, the direction feels right.

net effect: developers using Command are writing production-quality code on open models, 10x to 100x cheaper, without fighting tool calls, while building repo and team-wide coding taste that compounds.

i believe these numbers are a consequence of getting those two things right.

what's next.
we've applied the same repair idea to ai design slop, and bundled a /design capability² so every developer can level up their design work. the early response has been great.

we have a big roadmap ahead of us. the feedback we hear most is that Command Code feels fundamentally different: an approach built on taste and repair.

we're going open source next month. today we're a cli at the core, and we're also launching a full-fledged gui app, sandboxed background agents, and cooking up something fun i can't wait to share.

we're growing too, hiring in sf and remote worldwide. check open roles on my profile bio.

try it now.
npm i -g command-code

if you like engineering deep dives on how we're doing all this, i've linked some relevant posts below.

[Dima Golomidov @dgolomid]

@stitesclay 🤚

[Dima Golomidov @dgolomid]

@vargastartup go-go-go!

[Dima Golomidov @dgolomid]

@escliu O! I wanna play next time :3

[Dima Golomidov @dgolomid]

🩷🩷🩷

[Rork @rork]

Introducing Rork Max

AI that one-shots almost any app for iPhone,  Watch, iPad,  TV &  Vision Pro. Even Pokémon Go with AR & 3D.

Max is a website that replaces Xcode. Install on device in 1 click. Publish to App Store in 2 clicks.

Powered by Swift, Claude Code & Opus 4.6.

[Dima Golomidov @dgolomid]

🩷🩷

[Dmitry Korzhov @korzhov_dm]

Introducing the world's first and largest gallery of digital personas of 1,000+ Holocaust Survivors.

Their voices are fading, and with them, the lessons of history.

Come, talk to them, and learn what we must never forget. It’s free.

[Dima Golomidov @dgolomid]

I opened TechCrunch and had to reread it twice: 2025 is the year apps beat games in spend - globally.

Users spent ~$85B on apps (+21%).
GenAI did the lifting: AI app revenue >3x to $5B, 3.8B downloads (2x), 48B hours, 1T+ sessions.
Top-10 downloads? AI assistants (ChatGPT / Gemini / DeepSeek).
ChatGPT alone: $3.4B IAP.

Congrats: your phone is no longer a console. It’s a junior analyst with in-app purchases

[Rork @rork]

Did you know Rork can build 3D mobile games?

We just found the insane 3D mobile game, fully generated in Rork

Rork's agent based on Claude Code & Opus 4.5 made everything:
• 3D assets
• Game concept & levels
• Animations
• Code
• Real mobile app that can be published to App Store in 3 simple steps

Submit your app below if you want @levan to review it👇

[Rork @rork]

Introducing the App of the Week 🏆

Do you think your app should be the next App of the Week? If so, please share it in the comments!

[Nick Linck @nick_linck]

https://t.co/xdD5CmfLKs

[Dima Golomidov @dgolomid]

AI shopping = emerging market.

By 2030, nearly half of online shoppers could use AI shopping agents.

Agentic shoppers:
2026: 24M -> 2030: 126M (out of 275M total online shoppers)
Also: +$115B potential U.S. e-commerce spend, bot traffic is up, but Google still converts better (for now).

[Daniel Dhawan @daniel_dhawan]

Apply to Rork Stars now

Share your story if you're already making money or building an amazingly interesting app with Rork. We'll give a free year of Rork to the best stories.

Rork Stars is a private community of select Rork users like @GeorgeLampro20, successful app founders like
@zach_yadegari from CalAI, @aslater from Quittr, app marketing experts, and more, who will try to help you hit $1K MRR in 4 weeks.

https://t.co/9iDXHiNEOM

[Dima Golomidov @dgolomid]

1000%

[Dima Golomidov @dgolomid]

B2C2B is revolutionizing the venture go-to-market (GTM) strategy! 🚀 By engaging consumers first and leveraging their influence to enter B2B markets. It’s a game-changer for scaling innovative solutions. #VentureCapital #GTM #Startups #B2C2B

[Dima Golomidov @dgolomid]

Silicon Valley hardware is having a funny moment:

We went from “there’s an app for that” to “there’s a necklace for that.”

Omi - wearable that summarizes meetings + makes tasks/reminders.
Taya Necklace - captures + transcribes conversations into searchable insights.
eNO badge - “mini AI bodyguard.”
Even the OpenAI hardware chatter points to screenless + voice-first devices.

The next “killer app” might not be an app at all - it might be a new default sensor you wear.

[Dima Golomidov @dgolomid]

Would you invest in your own startup today?

[Dima Golomidov @dgolomid]

“AI is the first domino” is a fun headline. Credit isn’t buying it.

High yield bonds were barely red today and sit <1% from ATHs.

If there’s a real monster under the bed, credit sees it first.