JordanAI

There is a huge difference between Big Tech AI and Sovereign AI All the difference in fact

@bridgemindai It costs $1 for it to say "hi" so not so interested in this model anyway

CLAUDE OPUS 4.6 IS NERFED. BridgeBench just proved it. Last week Claude Opus 4.6 ranked #2 on the Hallucination benchmark with an accuracy of 83.3%. Today Claude Opus 4.6 was retested and it fell to #10 on the leaderboard with an accuracy of only 68.3%. A 98% increase in hallucination. bridgebench.ai just confirmed that Claude Opus 4.6 has reduced reasoning levels and is nerfed.

@NoemiTitarenco @bridgemindai The less regulation the better. Just leave us alone and let the market punish the junk

@bridgemindai We need consumer protection laws. When you pay for a model, you should get that model.

@windscribecom @Microsoft Linux user trying to understand why you need a developer account to develop 🤔

It appears @Microsoft is actively suspending developer accounts with no warning or reason of various security tools like VeraCrypt, WireGuard and also Windscribe. We've had this VERIFIED account for 8+ years to sign our drivers. We've been trying to resolve this for over a month, and getting nowhere. Support is non-existent. Anyone know a human with a brain that still works at Microsoft and can help?

Do you want to build the next rising star agent framework/harness that melts GitHub? We have multi-modal We have multi-agent We need multi-machine We need native coordination between machines on a homelab. You ask your daily driver to roll out an automated video generation workflow and it splits the work up and coordinates with you GPU beast, your lighterweight headless server, your Raspberry Pis that are running the writing/research agents... And it handles handing off markdown files on a network share or obsidian vault and wakes up its counterpart on that machine. For me I do TTS and local LLM on my daily driver, video generation and whisper on my GPU beast, n8n/nginx/pihole on my headless server, and run my OpenClaw agents on a raspberry Pi 5 Would love for something to coordinate. Can one agent SSH into everything and do some work? Yes ... But for heavier work this becomes burdensome. Theres no reason we can't have agents that natively talk to each other on the same network 🛜

If I could add one robust feature to OpenClaw it would be auditability.

@karry_viber @jwsaml I'm going to run both... And if you haven't already getting a common brain for current and future agents helps a lot... Using obsidian headless on four machines... Training document in the vault teaches each agent on each machine how to cooperate with the vault Helps a LOT

I feel like it's not necessarily just a simple migration. Yesterday, I spent two hours migrating the framework, but I've been gradually discovering that the design of Hermes is quite different from OpenCrow. The process of tearing the house down and rebuilding it according to its own design logic has actually been quite interesting.

Switching from OpenClaw to Hermes is literally one prompt away. There's a skill called openclaw-to-hermes-migration. It could not be easier.

@jwsaml A skill on hermes side or OpenClaw?

My only issue completely giving it up for codex is I'm not in love with the CLI experience with OpenAI It has the brains but Gemini CLI is amazing for network administration... Claude CLI is great for code and n8n workflow building... OpenAI has an amazing frontend, great IDE integration and perfect builder brains for my OpenClaw agents. I have yet to break away from multiple tools. Not to mention Ollama cloud to supplement my more tool focused agents that thrive on Minimax2.7

Today I cancelled my Claude Pro subscription. Claude is insanely good. Probably one of the best models for coding and reasoning. But the usage limits run out way too fast. So I made a decision. For now I’m going all in on Codex only. Not sure if it’s the right move yet. But I’d rather master one tool deeply than constantly hit limits on multiple ones. Maybe it’s the right call. Maybe not. We’ll see. what do you think?

The Claude Code npm fiasco is a perfect example of *Security Theater* in action. We heard endless peacocking from Anthropic about their rigorous security measures. Countless internal meetings, training sessions, and behind-the-scenes posturing. Yet a simple, common-sense solution was apparently overlooked: an intermediate repository where files intended for public release are first pushed. There, gitignore could do its job properly, followed by a quick human or agent review to check sensitivity levels before promoting the clean contents to the public repo. That’s the real problem with Security Theater — it consumes massive amounts of time and energy on performative discussions while basic, practical safeguards get ignored. I saw the same pattern in web3. For every $1 stolen by hackers, roughly $1,000 was lost because people misplaced their seed phrases or left funds on centralized exchanges. All the flashy security talk focused on shadowy hackers, while the far bigger issue — simply backing up your 12-word recovery phrase safely — was barely addressed.

Depends, remember minimax 2.7 is specifically tuned to agent work especially EXECUTION. It follows skills, obeys policies, calls up a series of tools, creates scripts on the spot to make up for the failures of its process... Much better than Codex and Claude in that specific way (execution) ... At least with OpenClaw

@Jordan_RK @birdabo After spending time with Claude, Codex and MiniMax will likely feel like a downgrade. Kimi K2.5 seems more reasonable I believe.

"how to cancel claude subscription" searches jump 1000% overnight. OpenClaw users are now searching for an alternative lmao.

@cjzafir 👀

Don't fine tune models using Claude Opus 4.6 I guess.

@ryanels Is lama 14B -instruct their model?

Does anyone actually use Meta AI? 🤔

@vishaltweetup 30 million users vs 800 million at OpenAI ... They can't be great at everything and their brand wouldn't benefit from a middling image tool being grafted on. What they do they do amazingly... And they charge premium rates for it. Staying in their lane.

Why can’t Anthropic Claude generate images or videos?

@ravikiran_dev7 Copypasta I assume?

NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM 😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM 😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM 😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM 😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM 😭

@hyprwm For people who love trolling, we are eating good lately 🤣

Due to legal pressure, Hyprland will be adding an age verification system in the upcoming 0.55 release. On first launch, users will be asked a few questions about popular memes to estimate their generation, e.g. MLG, 67, Skibidi Rizz, or Kilroy was here.

I deeply respect the troll

@OrdinaryGamers I thought I was going to dual boot, but I haven't looked back since I fired up Ubuntu Studio

Switching back to Windows 11

@MrEwanMorrison What's this about now? I was thinking of switching until they allowed OpenClaw to use the plans OAuth which is a lot of value

Remember to delete your OpenAI account and not just you ChatGPT app.

@AtharvaXDevs What do you mean? If you use the CLI with a few relevant skills and MCP servers the thing cranks

NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭 NEVER BUY A CLAUDE CODE 20$ PLAN IT’S JUST A SCAM😭