分类筛选
找到 1098 个 Demo
Introducing SERA-Crypto, a reasoning architecture...
Introducing SERA-Crypto, a reasoning architecture built from the ground up for state-of-the-art crypto research. SERA-Crypto is the #1 open-source agent on DMind and currently leads our live crypto benchmark, outperforming GPT-5, Grok 4, Gemini 2.5 Pro, and Perplexity Finance on real user queries, all with <45s end-to-end latency. How SERA works • Extracts true user intent • Resolves ambiguous tickers • Routes queries via semantic embeddings • Executes APIs + deep search in parallel • Intelligently loops on open-ended questions before synthesis • Produces grounded answers with sources The result: ReAct-level reasoning depth with predictable performance on live Web3 data, spanning DeFi, tokenomics, security, and smart contracts. Built for production crypto research, not demos. The full SERA framework, benchmarks, and technical deep dive will be open-sourced soon. @SentientAGI
Built my own multi-agent hive CLI using existing C...
Built my own multi-agent hive CLI using existing Claude Code, Codex & Gemini subs. Here's Claude Code orchestrating two Gemini 3 Pro agents for research Runs through bun + sqlite + tmux in bg Is this something you'd like to see me open source like my OpenSkills project? https://t.co/RzIyyttdKK
🔥 Insider Leak: The FierceFalcon checkpoint is no...
🔥 Insider Leak: The FierceFalcon checkpoint is now official for Gemini 3 Flash It features autonomous reasoning, tools integration a 1M-character context window and 65K output tokens Release could be as soon as Dec 18 #Gemini3 #Google #Gemini3flash #leak https://t.co/gHRyzFcc16
Trying to use AI to remove the weird highlights on...
Trying to use AI to remove the weird highlights on the faces in the reference pics. Didn't need to recompose, it kept the realistic details and matched the original 1:1. AI is just magic. #FC26 #Gemini3 #nanobanana https://t.co/8TZpnb4K9k
おやすみ（見知らぬ娘） #Gemini3 #Geminiai
おやすみ（見知らぬ娘） #Gemini3 #Geminiai https://t.co/FgjWO68kD3
遠い昔に「笑顔の方がいい」と誰かに言われた気がする #Gemini3 #Geminiai
遠い昔に「笑顔の方がいい」と誰かに言われた気がする #Gemini3 #Geminiai https://t.co/iLmSvqcEpV
22 people tried my new AI tool in the last 48 hour...
22 people tried my new AI tool in the last 48 hours — and it’s not even launched yet. I built a landing-page scoring engine powered by Gemini 3 Pro + n8n. And somehow… indie hackers, founders, and even teenagers in cyber cafés already started using it. If you care about conversion, clarity, or trust on your LP — try it now while it’s still free. Link in bio ⚡️ #buildinpublic
普通にぬいぐるみの写真撮って、 「テーブル背景の上で踊る」って指示しただけ。 凄いですよね🤔 #お...
普通にぬいぐるみの写真撮って、 「テーブル背景の上で踊る」って指示しただけ。 凄いですよね🤔 #おめシス #おめがレイ #Gemini3 https://t.co/TAOYEznxXt
Vibe coding robotics & engineering systems upd...
Vibe coding robotics & engineering systems update 2 Built another version of the robot arm simulation that can stack cubes to build walls. Designed with Nano Banana Pro, built with Gemini 3 https://t.co/ME4ka3gTuA
This guy literally built a full app UI using Gemin...
This guy literally built a full app UI using Gemini 2.5 in 3 minutes https://t.co/jirOXRLij7
Gemini 3 vs GPT 5.2 vs Claude 4.5 all tasked with...
Gemini 3 vs GPT 5.2 vs Claude 4.5 all tasked with building the same app → 1 prompt each ( same prompt ) → All built in Rork(.)com Comment below which one wins? https://t.co/4Tkwm00qab
Why stands out vs GPT-5 & Gemini 3: Because it’s...
Why @SentientAGI stands out vs GPT-5 & Gemini 3: Because it’s not a single model. It’s a reasoning + agent network. • combines multiple models instead of replacing them • agents execute decisions, not just generate text • transparent, open, and composable by design • built for real-world, high-stakes environments This is why Sentient is ahead.