




I AM
7.3K posts











i detach very fast. Anything i love extremely deeply I'm prepared to lose. It can break my heart but I have learned how to keep going. It can seem cold, but if it takes me off my pivot..l'll cut my finger off to save my hand. I can’t lose myself again, for anything. I'Il heal.


we all see how the team at Sentient has been shipping back to back, partnerships upon partnerships, improvements left and right, but it doesn’t even stop there. congrats to the team at @sentientAGI as they have been awarded “AI startup of the year” at Minsky awards 2025 - the largest and biggest AI conference in india. in case you don’t know what that means, this recognition reinforces sentient’s mission to make sure AI is open, decentralized, and community-owned, thereby challenging the centralization scheme of closed AI labs. the judges at Minsky awards obviously know something, they acknowledged sentient’s progress so far in democratizing access to advanced AI tools through ROMA (IYKYK 😒) you might think i’m saying all of these to sugarcoat things but it’s far more than that, so why not let’s talk some numbers .? > ROMA search achieved 81.7% accuracy on the FRAMES benchmark for multi-step reasoning, putting closed labs model (GPT-4o) in its place. just so you know, GPT-4o is 65.6% accurate on the FRAMES benchmark. while perplexity sonar is pro is only 44.4% accurate. 𝐭𝐡𝐚𝐭’𝐬 𝐧𝐨𝐭 𝐞𝐯𝐞𝐧 𝐚𝐯𝐞𝐫𝐚𝐠𝐞 𝐲𝐞𝐚 ? 𝐞𝐱𝐚𝐜𝐭𝐥𝐲 𝐦𝐲 𝐭𝐡𝐨𝐮𝐠𝐡𝐭𝐬 .!! > on another benchmark called the SEAL-0, the core of the 3 versions of SEAL-QA benchmark for complex reasoning, ROMA search achieved 45.6% accuracy. surpassing competitors like google’s gemini 2.5 pro, which is only 19.8% accurate on the SEAL-0 benchmark. if these improvements don’t tell you that sentient is cooking, and the team at sentient knows their stuff, and that they mean every word when they say they’re embarking on the journey to make AI open-source and decentralized, then i don’t know what else will tell you. anyway, i’ll see you on the next one as usual. gSenti frens 🎉





Current eligibility for Brevis Yapper Leaderboard: We're recognizing holders of our partner project tokens, showing how ZK attestation rewards genuine community participation and loyalty without sacrificing privacy. $LINEA ≥ 70000 $EDEN ≥ 7000 $USUAL ≥ 30000 $KAITO ≥ 1000 $mUSD ≥ 1000 Meet ONE to qualify.





If you’re Nigerian 🇳🇬 I respect your incomparable work ethic. I won’t pay you $5 / day. If you’re in my circle I’ll make you rich and pay you for your value, not from the country you’re from. Follow these exact steps (if you can’t follow these steps instantly disqualified). 1. DM me your skill set. 2. Leave a comment with #TeamWynn 3. Quote repost and tag Nigerian friends. - Wynn




in my previous thread about proof of attribution, we talked about the trifecta - the 3 core mission that the @gokiteAI team plans to achieve, which are; > decentralization of the AI economy > fair value flow > full transparency in the AI economy. 𝐛𝐮𝐭 𝐰𝐞 𝐧𝐞𝐞𝐝 𝐭𝐨 𝐚𝐬𝐤 𝐞𝐚𝐜𝐡 𝐨𝐭𝐡𝐞𝐫 𝐭𝐡𝐞 𝐫𝐢𝐠𝐡𝐭 𝐪𝐮𝐞𝐬𝐭𝐢𝐨𝐧𝐬. questions like; > how do they plan to achieve this ? > what does their tech stack looks like ? > who is it for ? … and a lot more. walk w me 🧵