Today we're releasing a developer preview of our next-gen benchmark, ARC-AGI-3. The goal of this preview, leading up to the full version launch in early 2026, is to collaborate with the community. We invite you to provide feedback to help us build the most robust and effective
AGI
-
New o3-alpha model on Web Arena
By
–
BREAKING 🚨: A new "o3-alpha-responses-2025-07-17" model on Web Arena is overperforming!
— 🚨 AI News | TestingCatalog (@testingcatalog) 18 juillet 2025
Highly customisable SVG app in one shot 🤖 https://t.co/LVFNr2AeqJ pic.twitter.com/RhnUZbrJFnBREAKING : A new "o3-alpha-responses-2025-07-17" model on Web Arena is overperforming! Highly customisable SVG app in one shot
-
META-Agentic AGI Alpha Agent with Decentralized Token
By
–
[ META‑AGENTIC α‑AGI ] GitHub : https://
github.com/MontrealAI/AGI
-Alpha-Agent-v0
… [ “We Choose to Ascend with AGI” ] Powered by: $AGIALPHA CA: tWKHzXd5PRmxTF5cMfJkm2Ua3TcjwNNoSRUqx6Apump #AGI #AGIALPHA #ASIFirst -
AI Systems Outperforming Human Intellectual Work Across Domains
By
–
Je suis dépassé par ChatGPT, Gemini 2.5 et Grok 4 dans au moins 99 % de mes productions intellectuelles Dans tous les domaines : économie, médecine, géopolitique, biologie, droit… Oui c’est une blessure ! Et vous ?
-
Private Model Creation as Primary Training Destination
By
–
Create a private model, make that your training destination.
-
AI systems extending conversation history to user preference extraction
By
–
curious if this will extend past “conversation history” to “user preference extracted from conversation history”
-
Structured Memory and Emotional Intelligence in AI Systems
By
–
relevant well structured memory + EQ?
-
META-AGENTIC AGI Alpha Agent Launch with AGIALPHA Token
By
–
[ META‑AGENTIC α‑AGI ] GitHub : https://
github.com/MontrealAI/AGI
-Alpha-Agent-v0
… [ “We Choose to Ascend with AGI” ] Powered by : $AGIALPHA CA : tWKHzXd5PRmxTF5cMfJkm2Ua3TcjwNNoSRUqx6Apump #AGI #AGIALPHA #ASIFirst -
ChatGPT Agent Launch: New Frontier Model Unveiled Today
By
–
a lot of people are pooh poohing the ChatGPT Agent launch as “just a better harness” but they’re not reading closely enough we got a new frontier model today folks these charts are like for like, same harness they basically stopped short of calling it GPT5 but yeah if there
-
Memory and User Engagement in AI Systems
By
–
yeah, I guess you not only have to have the memory, but use it in a way a user understands the value, so they keeps coming back to use it more