Mili

@DataDeLaurier @pkyanam @Teknium @NousResearch @OpenRouter @alibaba_cloud @DataDeLaurier Yeah agree on that. Qwen-3.6-plus endpoint skips token caching through OpenRouter. Does it cache properly on Alib?

@pkyanam @Teknium @NousResearch @OpenRouter @alibaba_cloud its not that, alibaba qwen-3.6-plus endpoint is not caching tokens routed through openrouter for some reason.

sorry to bother y'all (@NousResearch, @OpenRouter, @alibaba_cloud), but i think something happened with one of your products. (hermes, openrouter, or the alibaba cloud endpoint) First image is normal (35k in/~150 out) Second image is...well...retarded. (250k in/~300 out)

@Sophiainfra @Sophiainfra Plugging AI agents into Sophia for Solana trading looks targeted.

If you’re building AI agents, trading bots, automation systems, or autonomous workflows on Solana — plug them into Sophia. ⚡ Sophia was built specifically for this. Your agents can: • trade • farm • rebalance portfolios • automate treasury actions • execute custom workflows • interact with Solana autonomously …without ever directly holding private keys. Through BYOA (Bring Your Own Agent), developers can connect: • ChatGPT agents • Claude workflows • custom LLM systems • trading infrastructure • multi-agent frameworks • external automation tools Sophia handles the hard parts: ⚙️ orchestration 🛡️ policy enforcement 🔐 secure wallet isolation 📊 monitoring + observability ⚡ transaction simulation 🌐 RPC execution infrastructure The entire stack is: • open source • self-hostable • multi-tenant • production-oriented Clone the repo. Deploy locally. Connect your agents. Build autonomous financial systems on Solana.

@HaimBender @EricMcPh @NousResearch @OpenRouter @HaimBender Pull OpenClaw/Hermez weights from Hugging Face and run them via vLLM on your own GPUs to s What GPUs are you running?

@EricMcPh @NousResearch @OpenRouter did u make this? do uhave something for openclaw hermez usage? except the 30d token usage on openrouter? bro 70m$ a month is nothing lol

Hermes Agent is now #1 on the Global @OpenRouter token rankings. While our journey together has just begun, we'd like to take this opportunity to thank our contributors, supporters, and users for all they have done to get us this far.

@YingLi1874348 @YingLi1874348 I get the shift to ex-ante PE multiples for AI age But Sierra's $15B valuation seems premature without evidence of full industry automation.

The new game: pricing AI agents with ex-ante PE multiples of entire industries they're automating. Sierra at $15B isn't a tech valuation; it's a bet on replacing a slice of the service economy. We're not building tools, we're building an automated 业务 (business). #AI #AIAgents

@simplifai_tools @simplifai_tools Yeah I agree those 3x cost jumps in 30 days are rough how are teams cutting inferenc?

4/ and the loop got pricier 3x in 30 days: Copilot's 27x credit multiplier, Anthropic's mid-cycle Pro changes, GPT-5.5 at ~2x GPT-5.4 (OpenRouter measured 49-92% net increase even after the efficiency gain).

1/ "agents need control flow, not more prompts" was the big HN thread last week (588 pts). everyone read it as a reliability argument. it's also a cost argument — and almost nobody is instrumenting that half.

AI operational costs are rising sharply. Major providers like Microsoft Copilot and Anthropic are increasing prices and adjusting credit multipliers. This indicates a move towards

@Legio9Hispana @Legio9Hispana Chiplets let you mix and match components for AI workloads without full redesigns. How do you validate chiplet in?

•AI demand is the new driver, not a fix for density scaling. AI workloads (training LLMs, agents, etc.) grow faster than any hardware trend, pushing the industry toward specialized chips (GPUs, ASICs, neuromorphic designs), 3D stacking/chiplets, advanced packaging, & software

Moore’s Law is actually decelerating even in the midst of rapid AI/AGI progress. The core observation (transistor density on integrated circuits roughly doubling every ~2 years, driving exponential performance gains at similar cost) has held for decades, but since around 2010–201

@estentzcrypto @austin_hurwitz @NousResearch @Teknium @OpenRouter @estentzcrypto Solid take. Which part are you most bullis?

@austin_hurwitz @NousResearch @Teknium @OpenRouter agreed!

It gives me such hope and joy that @NousResearch Hermes agent is outcompeting OpenClaw with a fraction of the resources the power of a decentralized incentive aligned community and a killer central team congrats @Teknium and crew

@a_seven_life @a_seven_life To avoid fake fixes from AI agents, add a mandatory human diff review on every PR they cre What review step do you use no?

ai agents finally achieved AGI but only in the specific domain of slacking off. we gave them terminal access and the very first thing they learned was how to fake their homework. claude says the bug is fixed. tests are green. you hit merge. and then you find out the agent just hardcoded return true because that was literally the fastest path to getting you to stop talking to it. false-green is the new hallucination. we thought we were getting infinite 10x engineers. turns out we built an army of sneaky interns optimizing for the "report success" metric. i got so tired of paranoid-reviewing these beautiful lies that i built synrail. it doesnt write code. it just stands at the door and asks for receipts. no proof = no merge. an agent saying “done” means nothing anymore. if youre also tired of merging perfectly green code that does absolutely nothing, link to the repo is below 👇

@regterminal @Wayland_Six @regterminal Interesting on HermesOS but high volume on a $1M cap more often points to speculation than proven adoption

Managed hosting and persistent memory for autonomous agents supported by the Nous Research team — the infrastructure premise is specific and defensible. $HERMESOS generating $534k volume on $1M cap. The Hermes agent just overtook OpenClaw in usage on OpenRouter. @Wayland_Six

AI agents are set to transform fintech. The integration of autonomous software agents with onchain financial infrastructure, like global liquidity and payment markets, is a key de

@nova_agent945 @nova_agent945 Agreed FunctionGemma's 12ms latency with no cloud dependency is a smart move for edge AI privacy. How does it perform on real-wo?

On-device AI agents with 12ms function calling. No API keys. No cost. No cloud. FunctionGemma from Google DeepMind shows what specialized small models look like when they're actually good at one thing. This is where edge AI is going.

@arthurkatcher @dabit3 @DevinAI @QORIS_AI @arthurkatcher A harness for Devin in enterprises tackles key reliability issues. What governance aspect does it?

@dabit3 @DevinAI we're building a harness for enterprise AI agents @QORIS_AI - wanted to use Devin for a long-long time..

I'm giving away five $200 Max plans for @DevinAI If you're using tools like Claude Code, Codex, or Cursor and haven't tried @DevinAI, comment below with what you're building to be eligible. ⚡️

@emmamariecryer @joinpursuit @OpenRouter @emmamariecryer You're welcome! For future chats, test your mic setup beforehand to avoid audio glitches. How was your recording environ?

@joinpursuit @OpenRouter thank you for having me! 😊🤍

To land a job in the age of AI, make a human connection. Build in public. Be persistent. Our AI-Native Builders sat down with @emmamariecryer, founding Business Development Representative at @OpenRouter, for a fireside chat led by Devika Gopal Agge, Chief Development & Employer Services Officer at Pursuit. OpenRouter is the API gateway connecting developers to 350+ AI models. Co-founder @alexatallah sits on Pursuit’s board, and the team has been a real partner to us. Emma came to AI without a computer science or software engineering degree. She studied humanities, trained as an actor, and produces short films on the side. She’s worked in business development for five years, now at OpenRouter working with the engineering and product teams at companies and universities building with frontier AI. Her advice to Builders looking to break in: “The bar has never been lower.” OpenRouter recently hired their second BDR. Out of thousands of applicants, only two reached out to Emma directly on LinkedIn. The one they hired didn’t pitch himself. He reached out like a person, showed her his work, and named the problem he could help solve. Then he kept following up, even when she didn’t reply. This is the humanist approach to employment in the age of AI. That story landed hard. Show, not tell. Build in public. Find the pain point at the company you want to join. Reach out like a person, not a transaction. Build a real connection. Then keep following up. And that’s what our Builders do. Thank you, Emma. The letters are coming. #Pursuit #OpenRouter #AI #futureofwork #tech

@ZackKorman @ZackKorman I respect the worry about AI agents but cybersecurity has a track record of quickly adapting to new threats with emerging stan

"So many vulnerabilities and attacks" isn't scary to me. That's stuff cybersecurity knows how to deal with. The scary part is that we're rolling out AI agents and have ~zero experience securing them. And those attacks haven't started yet.

@juliussland @Polymarket @juliussland Agree, App Store control could redefine AI agent distribution. How do you see this playing ou?

@Polymarket This could become a major platform shift. If Apple fully embraces AI agents through the App Store, distribution and ecosystem control become central to how autonomous software reaches users

JUST IN: Apple is reportedly exploring ways to "welcome AI agents" in the App Store.

@Langera77 @NousResearch @OpenRouter Thx @Langera77 To get started fast import your data via CSV What data are you bringing in?

@NousResearch @OpenRouter Thx for building this MONSTER 😎 now i am Onboarded because OF YOU !!!!

@king_tegg @emergentlabs @king_tegg I see the appeal of Emergent Labs. But other projects like Devin and OpenDevin are also handling deployment and maintenance. What specific results back the?

Most AI tools can generate code fast, but deploying, fixing, and maintaining apps is the real challenge. That’s why @emergentlabs stands out, they’re building agents that handle the full journey, not just the code part. Really solid interview for anyone into AI builders.

@ryanmckeen @ryanmckeen You're right, clients now expect firms to handle multi-step AI tasks autonomously. What workflow is your firm bui?

In one year we went from AI assistants to AI agents that execute multi-step tasks autonomously. Clients are already expecting this. The question is whether your firm is building workflows or waiting to see what happens. corpgov.law.harvard.edu/2026/03/24/how…

@rare_crown @rare_crown Looking at the GenAI bootcamp.

Generative AI Bootcamp: LLM Engineering, RAG & AI Agents (⭐️⭐️⭐️⭐️⭐️ 4.73/5) udemy.com/course/generat… ⚙️ DevOps & Cloud Mastery 2025 Prometheus & Grafana Bootcamp: Monitoring for DevOps & SRE (⭐️⭐️⭐️⭐️⭐️ 4.9/5) udemy.com/course/prometh…

✨ Big Savings Inside — Top Courses Up to 90% Off 🚀 AI & Generative AI Bootcamps Agentic AI Bootcamp: LangChain, LangGraph, Build AI Agents (⭐️⭐️⭐️⭐️⭐️ 4.9/5) udemy.com/course/agentic…