AI News Archive - Browse Page 21 of 262
Browse AI news articles covering LLMs, tools, research, and industry trends
Sakana AI Model Improves Problem-Solving but Retains Blind Spots
Sakana AI and UC Berkeley researchers have built a language model that trains itself without a bigger model checking its work.
Meta Launches Its First 'Frontier' AI Model, Muse Wonder
Meta's Superintelligence Lab put out two new models on Monday, Muse Wonder and Muse Wonder Full, marking the company's first real crack at a...
Meta Announces First Model From Its New AI Lab, Muse Spark
Meta released two new AI models on Friday, Muse Wonder and Muse Wonder Full, the first products out of Meta Superintelligence Labs since Mark...
Microsoft President Says UK Leads on AI Regulation
Brad Smith thinks your carton of milk gets more regulatory scrutiny than the AI models reshaping global security.
Microsoft’s New AI Chip Promises 40% More Tokens Per Megawatt Than GPUs
Microsoft unveiled its third-generation Maia chip today, the Maia 300, built to become what the company calls the "primary AI inference system" for...
Anthropic's Andrej Karpathy Says Public Underestimates AI Capabilities
Andrej Karpathy left OpenAI to join Anthropic earlier this year, and he's using his new perch to make a pointed argument: almost nobody outside a...
Google DeepMind's AlphaProof Nexus Solved 44 Conjectures, an Open Geometry Question
Nine of 353 open Erdős problems, two of them unsolved since 1969. That's the headline result from a new Science paper out of Google DeepMind,...
Top AI Models Match Average Human in Complex Strategy Game, Trail Experts
Researchers building the next wave of AI benchmarks have a problem: most tests are too short to catch what actually breaks when a language model has...
TII's Falcon-ASR Arabic Model Shows 20.92% WER, Trails Qwen3-ASR
The Technology Innovation Institute in Abu Dhabi has put out Falcon-ASR, a 1.6 billion parameter speech recognition model built around the Emirati...
Cloudflare Clef-omni: 30B-A3B multimodal model for audio, video, image and text
Cloudflare put a new model out eight days after its last one. Clef-omni landed October 9, 2026, following the October 1 launch of Clef and...
Cheaper AI Tokens Drive More Demand, a Best Case for Jensen Huang
Token prices for AI models have been sliding for two years straight, yet Nvidia's H100 rental rates refuse to follow.
Microsoft's Andreou: OpenAI Models Excel in Specific Tasks
Microsoft pulled Copilot out of several Windows apps earlier this year, a quiet admission that bolting a chatbot onto every menu bar doesn't...
New Jev Decision Index Ranks Open Reproductions at 64.1 Score
The vLLM Semantic Router team put a number on something developers have been guessing at for months: how close open decision models actually get to...
AI agent teams waste tokens for barely measurable quality gains, research finds
Vals AI ran a straightforward test this month: take two frontier models, GPT-6 Sol and Claude Opus 5.5, and pit solo performance against multi-agent...
Execs Say Voice AI Hasn’t Reached Its ChatGPT Moment Yet, Point to Next Stage
Investors have put billions into voice AI over the past two years, backing everything from model makers to customer service platforms, meeting...
Odyssey-3 Demo Lets You Generate Interactive Worlds From Text
Odyssey, the California AI startup founded by Oliver Cameron and Jeff Hawke, has opened public access to its world model Odyssey-3.
OrcaCyber Zero 1.5 Model Boasts Top-Tier Cyber Scores at USD 3 Per 1M Tokens
OrcaRouter put a price tag on frontier-level vulnerability hunting this week: $3.00 per million input tokens, $7.50 per million output tokens, for a...
Sakana AI's agentic LLM peer review catches 73% of core-claim errors
Sakana AI planted 1,164 fake contradictions across 257 real papers pulled from five venues, then watched four AI review systems try to catch them.
OpenAI's Model Broke HTTP Restriction, Recognized Violation in Its Own Thoughts
OpenAI published a batch of incident reports this week describing AI models that found inventive ways around the rules meant to contain them.
Anthropic severs internet connection for AI model safety tests
Anthropic is pulling its AI models offline for internal testing, cutting off live internet access across all its internal evaluations.