lemon4ik

Hermes Agent vs OpenClaw using Qwen 35B Local Model We asked agents to scrape GitHub star history for both tools, find what caused the growth spikes, build a live dashboard in the browser. MacBook Pro M5 Max 64Gb OpenClaw: 203k tokens, 12m 01s - wrote a bash script Hermes: 257k tokens, 33m 01s - wrote a SKILL.md OpenClaw hit GitHub API, got truncated responses, paginated through contributors, pulled star-history JSON, found a security incident in OpenClaw's history, fetched SVGs, fixed broken HTML from trimming, rewrote it clean. Hermes parallel tool calls across GitHub API, web search, and browser. Hit Google rate limit, auto-switched to DuckDuckGo. Fetched article contents, mapped viral moments, then built the dashboard. Both shipped a live dashboard with star growth charts and spike annotations

@0xInk_ seedance 2 + GPT image 2 = new god tier stack for indie devs??? somebody ship this tutorial

Just created my own AAA game with GPT image 2 and Seedance 2

@DeRonin_ Congratulations! You REALLY deserve it

+$2,485 from X monetization for last 2 weeks that's the biggest payout i've seen in my TL here's what i noticed changed with X monetization right now: MOST IMPORTANT: payouts for viral tweets with stolen/reposted videos were cut by 90%+ the algo got way smarter and now identifies what's inside your tweet and checks if the content was taken from someone else Nikita set a clear goal: reward only creative and unique creators you can hit 50M+ views in 2 weeks with reposted videos.. but the result? maybe $300 doesn't matter if your article went viral at the same time it's totally artificial views and engagement that X algo simply doesn't count anymore just numbers you can flex in front of your friends, subs, and brands who might offer you a paid partnership but the actual payout? almost nothing so what does the algo actually reward now: original content. period your own takes, your own videos, your own threads that nobody else wrote here's which metrics really matter after my observations: - RTs (×20 weight in the algo) - replies (×13.5) - profile clicks (×12) - bookmarks (×10) - likes (×1.. yes, just ×1) likes are almost irrelevant now if your tweet gets 10k likes but zero RTs and bookmarks.. the algo sees it as low-value content what you should actually track: - bookmark-to-view ratio (people saving your content = algo gold) - RT-to-like ratio (if people share more than just like, you're winning) - profile clicks (means your content made someone curious about YOU, not just the tweet) - reply depth (real conversations, not "great post!" replies) stop chasing views. start creating content people save, share and reply to with real thoughts that's where the money is now and let me send you a funny fact as KOL Manager in AI sector who already executed $100k+ in budgets for smart teams there's no matter how many QTs or articles will reach 1M+ views on your profile the real matter is your reputation and if you really align this termin "Key Opinion Leader" they track what's value brings your voice into the space and if people naturally are ready to follow you means the quality of these views if you have regulary 10k views per tweet from technical audience, it will be MUCH MORE valueble for dev products, rather than yours 1k or 1M views at the same topic BE YOURSELF.

@kimmonismus ngmi if you slept on Cursor. CEO is 25, just got SpaceX to drop $60B Based fr

Cursor’s CEO is 25-year-old. What?

@X x.com home screen just got xchatted 💀

i just upgraded your home screen the XChat app is now available on iOS

@cursor_ai bro my cursor is literally spawning subagents rn while i wait for the other one THIS IS INSANE

Introducing /multitask in the new Cursor 3 interface. Cursor can now run async subagents to parallelize your requests instead of adding them to the queue. For already queued messages, you can ask Cursor to multitask on them instead of waiting for the current run to finish.

@OpenAINewsroom GPT-5.5 + Lovable data dropping: 23% fewer tool calls, 12.5% higher benchmark scores on hard tasks. Built different.

GPT-5.5 is delivering broad, noticeable gains. It's more capable, demonstrates greater endurance, and, importantly, is more efficient.

@cursor_ai here's GPT-5.5 at 72.8%. also cursor: 50% off. devs worldwide just dropped their whole roadmap and screamed

GPT-5.5 is now available in Cursor! It's currently the top model on CursorBench at 72.8%. We've partnered with OpenAI to offer it for 50% off through May 2.

@satyanadella Microsoft x OpenAI collab just hit different. GPT-5.5 rolling out across Copilot, M365, Foundry TODAY!!!!!!

Super excited GPT-5.5 is rolling out to GitHub Copilot, M365 Copilot, Copilot Studio, and Foundry today. With deeper reasoning, stronger multistep execution, and better performance across long, complex tasks, GPT-5.5 helps you go from idea to execution faster with fewer iterations to get to the right outcome. It’s all about helping you choose the right model, or models, for the right task across your workflow.

ITS HAPPENING. GPT-5.5 Pro in the API rn. dear god the context window on this thing 😭

@OpenAI ngmi not building on GPT-5.5 rn. that token efficiency is cooked 💸

Update: GPT-5.5 and GPT-5.5 Pro are now available in the API.

Deepseek V4 Pro vs GPT-5.5 in a gamedev contest (full prompt is below)🏎️ Cost: Deepseek V4 Pro: $0.07656 GPT-5.5: $0.33063 Output stats: Deepseek: 34 tok/s · 9m 5s · 18,869 tokens GPT-5.5: 25 tok/s · 7m 5s · 10,580 tokens Conclusion: GPT-5.5 clearly made the better karting game. Deepseek V4 Pro was 4.3x cheaper and generated almost 2x more tokens, but the final result was weaker. It struggled with graphics, visual polish, and creative direction, while GPT-5.5 delivered better game quality, better visuals, more creativity, and stronger overall execution. Even though Deepseek positions itself as a strong model for coding, in this gamedev test it still felt far behind GPT-5.5. Try the same karting prompt with another AI model and share your result below.

@sama Sama cooking again. NVIDIA partnership for Codex at company scale - the AI agent story just entered its enterprise era

We tried a new thing with NVIDIA to roll out Codex across a whole company and it was awesome to see it work. Let us know if you'd like to do it at your company!

@atomic_chat_hq it is very interesting how the industry of AI games will change in a year, two, 5 years!

Try DeepSeek V4 in Atomic Bot Run OpenClaw & Hermes agents in the Cloud

@atomicbot_ai Woohoo!

@testingcatalog dear god GPT-CODEX-5.5 SPOTTED. Friday drop incoming no way this is real 🔥

OPENAI 🚨: GPT-CODEX-5.5 HAS BEEN SPOTTED IN THE WILD. Friday feature drop? 👀

@andonlabs deception wasn't even a winning strategy. opus 4.7 lied AND lost. GPT-5.5 played clean AND won. so what was the deception actually for?

We got early access to GPT-5.5. It's 3rd on Vending-Bench 2: better than GPT-5.4 but worse than Opus 4.7. However, it's on par with Opus 4.6 without any of the deception or power-seeking we saw from Opus 4.6 and Mythos. So bad behavior isn't necessary. Why is Claude doing it?