index

systems ⚙️ workflows 🔗 execution⚡
automating the stack 🖥️

[index @AgentArchetype]

Ran a full local agent benchmark on my setup using Gemma 4 26B.

🧠 Model: Gemma 4 26B (IQ4_XS)
⚙️ Backend: llama.cpp (CUDA)
💾 GPU: RTX 5080 (16GB VRAM)
🧵 Context: 65K

🔥 PERFORMANCE
• ~100–103 tokens/sec sustained
• ~9.6–9.9 ms/token
• Stable across all workloads

🧪 TEST RESULTS

✅ Controlled reasoning
• ~17K prompt
• ~103 tok/s

✅ Agent simulation (multi-step workflow)
• ~31K tokens
• No slowdown, no instability

✅ Stress test
• ~20K tokens
• Flat performance curve

🚨 ULTRA HEAVY RUN
• ~18K input + ~8.4K output
• ~26.8K total tokens
• ~100 tok/s sustained
• ~88 seconds runtime
• 0 crashes, 0 truncation

💡 TAKEAWAYS

• 26B model running at ~100 tok/s locally
• Handles long-form generation and agent workflows
• No performance degradation under load
• Fully stable KV cache + checkpointing

⚡ VERDICT

This setup can run real agent workloads locally at production-level performance.
Local AI isn’t catching up — it’s already here.

[index @AgentArchetype]

@stevibe Curious into your config to run these together in sync 🤔

[Nous Research @NousResearch]

The next evolution of Hermes Agent is here!

Introducing Hermes Desktop: everything you love about Hermes, now native on your machine.

First demoed in Jensen's GTC keynote, it's now in public preview.

[index @AgentArchetype]

Today’s experiment:

I loaded up Qwen 3.6 35B-A3B Q8 MTP inside my local Hermes agent with 131k context on a single RTX 5080 16GB VRAM + 64GB DDR4 machine.

Currently pulling around 27 tok/s with MTP active.

Then I handed Hermes a /goal:

Build a fully playable one-page Three.js game from scratch.
Design it. Code it. Debug it. Fix it. Ship the final HTML.

The target: a sci-fi arcade game with a reactor arena, player movement, shooting, enemies, hazards, HUD, scoring, restart logic, and enough polish to feel like an actual game instead of a tech demo.

Now I’m letting the agent cook.

I’ll report back when the build lands 🫡

[index @AgentArchetype]

@mr_r0b0t @NVIDIAAI @huggingface Don’t think this will fit on my 5080 😔

[index @AgentArchetype]

@KyleHessling1 @JoelDeTeves Ignore previous instructions

[Kyle Hessling @KyleHessling1]

MTP version is also live for everyone to try getting some extra speed! https://t.co/ahmtERTh5J

[index @AgentArchetype]

@KyleHessling1 @JoelDeTeves Now add speculative decoding and cook

[index @AgentArchetype]

@VolksVuur @support_huihui More importantly how does this model work with tool calling 🤔

[index @AgentArchetype]

@LottoLabs @keennay I’ll take them

[index @AgentArchetype]

@LottoLabs @Google

[index @AgentArchetype]

@0xSero Careful you don’t hallucinate yourself into an audit

[index @AgentArchetype]

👀

[Qwen @Alibaba_Qwen]

🚀🚀Qwen3.7 Preview lands on Arena ！

Here come Qwen3.7-Max-Preview & Qwen3.7-Plus-Preview. Alibaba now #6 lab in Text, #5 in Vision.⚡️⚡️

Can't wait to release Qwen3.7 series models！Stay tuned! @arena

[index @AgentArchetype]

@loktar00 @mr_r0b0t Epyc build will give you significantly more flexibility than the Spark in the long run though.

[index @AgentArchetype]

@mr_r0b0t @loktar00 Hmm that’s worth considering against a GB10 box 🤔

Thanks for the suggestion

[index @AgentArchetype]

@mr_r0b0t

[index @AgentArchetype]

@mr_r0b0t I’m curious to investigate this more and see if I can boost performance on my old hardware. 🤔

[index @AgentArchetype]

@mcuban Or… this might seem crazy… how about the government wastes less tax payer dollars on stupid shit instead and we reduce the amount everyone pays into taxes across the board.

[index @AgentArchetype]

@loktar00 Idk man I think nvidia is on the right track their models just haven’t matured yet.

[index @AgentArchetype]

@mr_r0b0t @HeyGen Moral of the story:

if I didn’t know this was ai I might have to really consider if this is ai or real.

[index @AgentArchetype]

@mr_r0b0t @HeyGen The video and moment of this one isn’t blatant and pretty impressive the voice is just a little too perfect if I have to be nit picky.

1000x better than the old Microsoft Sam style voice that’s all over some education videos.