Sebastian Szewczyk

[Jakub Chmura @_jakubchmura]

ruNNtime v0.2.0 just got a huge performance boost 🚀
We integrated a lot of findings from @huggingface kernels library into ruNNtime, speeding inference up significantly. huge thanks for your help on this @nicodotdev @xenovacom 🤗

[Sebastian Szewczyk @sebscode]

My friend spent years studying architecture.
I passed Claude Certified Architect - Foundations exam.
I'll be calling her "a fellow architect" until she asks me to stop.

[Software Mansion @swmansion]

Software Mansion is now a Select partner in the Claude Partner Network 🎉

If you're looking for an AI-first team that builds with the latest tools, let's talk.

More in our blog post: https://t.co/HqGTmL2O3D

@AnthropicAI @claudeai

[Software Mansion @swmansion]

What it takes to win the Close the Loop challenge? 🏆
We're judging in three categories:
🏅 Best Overall: The most impressive demo end-to-end.
🌐 Best 3d: The most creative use of 3D graphics (we may give extra points for using TypeGPU 👀)
🤖 Least Slop: The cleanest agent run with the fewest retries, hallucinations or unnecessary steps.

Enter today: https://t.co/16yNuFctQx

[Sebastian Szewczyk @sebscode]

@poteto @alxfazio When Cloud Agents would be available with SuperGrok subscription?

[Software Mansion @swmansion]

We're launching the Close The Loop challenge today! 🏆
> build an app
> let Argent test it
> record the demo and post with #CloseTheLoop hashtag, and you're in!

Show us how you use Argent and win prizes along the way: https://t.co/aQpb7OX8Bp

[Sebastian Szewczyk @sebscode]

@tomekzaw_ fetch alternative?

[Sebastian Szewczyk @sebscode]

@thymikee Congrats! I build a lot of small hobby games and sites with coding agents. I'd rebuild one of my recent ones with Apex from the same prompt and see how it compares.

[Software Mansion @swmansion]

Introducing Argent Cloud

Cloud Simulator Infrastructure for AI Agents

[Konrad Reczko @reczko_konrad]

You can do some pretty fun realtime VFX once the depth output is just another (Type)GPU buffer :D

I basically yoinked my old 3D fire sim (that is actually a fluid sim but do not tell anyone) and plugged it into the depth pipeline together with hand tracking

Everything here is built from scratch in TypeGPU, so inference, simulation and rendering can share GPU resources and interleave passes directly. That’s 2 NN inferences, a full 3D fluid sim, relighting and rendering in one command encoder, still under 16 ms with no CPU sync in between

[Sebastian Szewczyk @sebscode]

@noahzweben Hey Noah, thanks for sharing the good news! Are the terms for that promotional period described in detail somewhere?

[Claude @claudeai]

Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.

Its capabilities exceed those of any model we’ve ever made generally available.

[Software Mansion @swmansion]

⚛️ Learn how Argent improved AI workflows at Software Mansion, helping our engineers cut re-renders and render times on real production apps.

[Sebastian Szewczyk @sebscode]

11/ The capability curve (Alex Albert)

SWE-Bench Verified: 62% (Sonnet 3.7) → 87% (Opus 4.7).

Each generation, *simplify* — strip scaffolding built around old models.

'Boost performance by removing instead of adding.'

https://t.co/WsOnrqjmKd

[Sebastian Szewczyk @sebscode]

Code with Claude 2026 wrapped this week. All 11 talks, with links and a one-line takeaway each.

[Sebastian Szewczyk @sebscode]

10/ Getting more out of the Claude Platform (Brad Abrams)

Four core techniques: prompt caching (90% input discount; cached tokens don't count vs. rate limits), Tool Search Tool (lazy-load tools), programmatic tool calling, compaction.

https://t.co/Dxw7OOJs7v