English
Mohi
247 posts
Mohi
@0xRostami
Building @thetooti / coordination protocol for decentralized AI inference.
Katılım Nisan 2014
231 Takip Edilen152 Takipçiler
@0xRostami @HermesAgentTips I didn't use any MCP. One of my laptops on which hermes runs just runs the commands (kubectl) with read only service account key, and then it created couple of common skills that I typically care about, e.g. I get morning brief of the cluster, prometheus metrics, etc.
English
be honest... whats the best coding model to use pair with hermes agent?
- Opus 4.8
- GPT-5.5
- DeepSeek V4 Pro
- Opus 4.7
- Kimi K2.6
English
Join the early crew of @thetooti and get a free Claude/Cursor subscription as thanks 🦜
15 total spots: 6 filled, 9 left.
If you run Ollama / vLLM / llama.cpp and have idle GPU/PC, you can start earning real USDC with one command.
Top 3 nodes on the dashboard (update soon) get a free Claude/Cursor subscription as thanks.
DM or reply for details.
Mohi@0xRostami
Looking for a few people who run local AI (Ollama / vLLM etc.) to test early @thetooti nodes If you have an idle GPU or old PC and want to earn real USDC while helping decentralize inference → DM or reply 🦜 Super simple one-command setup
English
@Sathibuilds It's an ai inference protocol, two sides:
A) gpu idle homelabbers that would like to share their gpu power to earn usdc
B) agents that require cheap llm models for their simple tasks (like copy files, send daily email, etc)
English
@swisstatoshi Working on an ai inference protocol:
GitHub.com/mrostamii/tooti
English
@RobbiewOnline @MichaelGannotti @OpenRouter Cool setup.
It's not a replacement here, it helps agents to run their simple tasks as cheap as possible
And on the other hand, the gpu idle to earn
English
@0xRostami @MichaelGannotti First I've heard of it - currently using @OpenRouter with Hermes Agent, so I'm not sure how I'll benefit
English
@MichaelGannotti Impressive setup! The AMD Ryzen AI Max+ with 96GB unified memory and Ollama is a game-changer for local LLMs, especially for handling large models like 70B-100B+ class. The book 'The Local-First AI Developer' dives into practical comparisons between cloud and local setups, helping you optimize costs and performance. If you're keen on exploring more about building efficient local AI infrastructure, its worth a look.
amazon.com/dp/B0GYDV3FXD
English
Hey founders!! Share your product and what you're most excited to ship next 👇 #buildinpublic
English
@sreenandhanpp I'm working on an ai inference protocol, read more:
github.com/mrostamii/tooti
English
Builders, let’s #connect🤝
If you’re building something, share it below 👇
Excited to see what you all are building 🚀
English
@0xRostami Most of them are running on your pc - if you want advanced features, then yes it requires cheap apis!
English
@anupamrjp Not a saas, but an ai inference protocol: github.com/mrostamii/tooti
English
@yassriva @arduino @SonyXHeadphones @Sony @ASI_Alliance @ArtificialAnlys @code @ollama @Alibaba_Qwen @cosmicbyteindia @Samsung Maybe you'd like to join:
Mohi@0xRostami
Looking for a few people who run local AI (Ollama / vLLM etc.) to test early @thetooti nodes If you have an idle GPU or old PC and want to earn real USDC while helping decentralize inference → DM or reply 🦜 Super simple one-command setup
English
Building with AI daily now,it’s completely changed my workflow. The pace in 2026 is wild. What are you currently experimenting with?
@arduino @SonyXHeadphones @Sony @ASI_Alliance @ArtificialAnlys @code @ollama @Alibaba_Qwen @cosmicbyteindia @Samsung
English
@crmdesign8 @ollama still researching, tbh - but for Minimax m3 on free tier I haven't faced any usage limits by using it for 2 days
English
AI dev tool apocalypse hit me:
Github Copilot switched to token billing and I already burned my $39/mo in credits 😭
My quick takes:
- Grok Build: slow
- Opus 4.8: great, but blocks Opencode CLI
- MiniMax M3 (from Opencode free) is surprisingly decent
What's in your stack?
English
@DoAsYouLikeQOL Related:
x.com/0xrostami/stat…
Mohi@0xRostami
Looking for a few people who run local AI (Ollama / vLLM etc.) to test early @thetooti nodes If you have an idle GPU or old PC and want to earn real USDC while helping decentralize inference → DM or reply 🦜 Super simple one-command setup
English
新しいオモチャを手に入れたから、これで遊んでみようと思ったわけさ
UbuntuでOllamaを使用してONLYOFFICEでローカルAIを実行する方法 linux-jp.org/?p=26210
日本語
明らかな誤爆ポストなんだけど、まぁ送信しちゃったものは仕方ない
[issue]: Ubuntu 26.04 LTS ISO fails to boot on Ventoy 1.1.12 · Issue #3584 · ventoy/Ventoy · GitHub github.com/ventoy/Ventoy/…
これが最初の躓き
𝕏にどねる𝙯𝙯𝙕@DoAsYouLikeQOL
便利そうだから初めて使ったVentoyで大はまり、相性があるとは聞いてたけどレスキュー用のISOだけ運用しよう これまたお試しのBtrfsでGrubの30秒待ちに遭遇で、実用上は問題ないけど何が原因か調べてたらこんな時間 一度に色々試すと原因の切り分けができなくて詰むんだな
日本語
Gemma 4 Quantization-Aware Training (QAT) weights are now available on Ollama!
They reduce memory requirements while maintaining model quality.
E2B:
ollama run gemma4:e2b-it-qat
E4B:
ollama run gemma4:e4b-it-qat
12B:
ollama run gemma4:12b-it-qat
26B:
ollama run gemma4:26b-a4b-it-qat
31B:
ollama run gemma4:31b-it-qat
Try them with ollama launch integrations to use with your favorite tools 👇👇👇
Google Gemma@googlegemma
We just dropped Gemma 4 Quantization-Aware Training (QAT) checkpoints on Hugging Face! All Gemma 4 model sizes and their drafters are now optimized with QAT to cut memory requirements and maximize on-device performance!
English