Google released Gemma 4, a family of four open multimodal models under Apache 2.0. The 31B dense model scores 89.2% on AIME 2026, quadrupling Gemma 3. For creators and developers building on open models, this changes everything.
Liquid AI releases LFM2.5-350M, a 350M parameter open-weight model that runs AI agents on mobile GPUs in just 81MB with 32k context window.
A misconfigured data cache exposed Anthropic's biggest secret: Claude Mythos, an AI model the company says represents a step change over Claude Opus 4.6 in coding, reasoning, and cybersecurity.
Anthropic accidentally exposed Claude Mythos, described as its most powerful model ever, through an unsecured data cache. The model shows major advances in coding, reasoning, and cybersecurity.
Mamba-3 matches transformer quality at 1.5B parameters while running inference 7x faster. Three technical innovations close the SSM capability gap.
Meta has delayed the launch of its next-generation AI model, codenamed "Avocado," from March to at least May 2026. Internal testing revealed the model falls short of competitors from Google and OpenAI, landing somewhere between Google's Gemini 2.5 and Gemini 3 in performance
Google released Gemini 3.1 Pro on February 19, 2026, reporting more than double the reasoning performance of its predecessor on key benchmarks. The model scored 77.1% on ARC-AGI-2, a test designed to measure general intelligence capabilities