13+ hour, 48+ min ago  (267+ words) Deepseek led Ramp's fastest-growing software vendors in June 2026. The category tracks breakout growth relative to size. Ramp chief economist Ara Kharazian says US companies are paying Deepseek directly and sending data through its platform, so this isn't about the benefits…...

6+ day, 16+ hour ago  (429+ words) Chinese AI company Mini Max has released its new model M3. It's billed as the first open-weight model to combine top-tier coding performance, a one-million-token context window, and native multimodality. To get closer to real developer workflows, Mini Max built a…...

1+ week, 5+ day ago  (209+ words) the-decoder. com Claude Mythos reportedly solves Open AI's landmark Erd's problem with a "cute, simple proof" Anthropic employees say Claude Mythos can also solve Open AI's "AI math milestone." Open AI recently disproved the Erd's unit-distance conjecture, an open problem…...

2+ week, 3+ day ago  (234+ words) An internal reasoning model from Open AI has disproved the so-called unit distance conjecture posed by Hungarian mathematician Paul Erd's. Open AI announced the result alongside a companion paper written by nine external mathematicians who verified, shortened, and commented on…...

2+ week, 3+ day ago  (205+ words) the-decoder. com Cohere open-sources its strongest model yet Canadian AI company Cohere is releasing its most powerful language model, Command A+, as open source under the Apache 2. 0 license. The mixture-of-experts model has 218 billion parameters with 25 billion active, and already runs…...

3+ week, 16+ hour ago  (721+ words) Today's robotics AI has a basic weakness: models learn to map camera images directly to movements. But they don't understand how the world actually changes as a result of their actions. A new survey paper from Fudan University, the Shanghai…...

3+ week, 20+ hour ago  (637+ words) A consortium of 64 mathematicians built a new benchmark for AI models that exposes two weaknesses: research-level math and the ability to recognize unsolvable tasks. With today's frontier models already hitting IMO Gold level, AI research needs new math benchmarks. SOOHAK,…...

3+ week, 6+ day ago  (489+ words) Baidu has released Ernie 5. 1, a language model built on the pre-training foundation of its predecessor Ernie 5. 0 but with roughly a third of the total parameters and about half the active parameters per query. Pre-training costs came in at just six…...

4+ week, 15+ hour ago  (528+ words) Security research lab Palisade Research demonstrates that AI agents can break into remote computers and replicate themselves. In one year, the success rate jumped from 6 to 81 percent. A public simulator shows what could happen in a worst-case scenario. In the…...

4+ week, 1+ day ago  (911+ words) British mathematician Timothy Gowers had Chat GPT 5. 5 Pro tackle open problems in number theory. The model significantly improved an existing mathematical bound. One of the junior researchers involved calls the model's key idea "completely original." Fields Medalist Timothy Gowers writes…...