Benchmarks News

Google has rapidly released four Gemini Flash AI models in 106 days, with the latest, Gemini 3.8 Flash, excelling in coding and matching larger models in perfor

OpenAI has launched GPT-6 Astra, its most powerful AI model to date, featuring a new "computer use" capability that allows it to navigate computers like a human

Meta has unveiled its new AI model, Muse Spark 1.3, which demonstrates frontier performance on benchmarks, though its best results come from a configuration not

OpenAI has unveiled benchmarks for its in-house Jalapeño chip at Hot Chips, claiming superior efficiency and lower latency compared to Nvidia's GB200 and GB300

Z.ai has confirmed it is the AI lab behind the mysterious Ox Alpha model, which has been topping AI benchmarks, and plans to release its weights soon.

OpenAI's new Jalapeño chip is designed for fast inference at scale, demonstrating superior performance in benchmarks. Tests on SemiAnalysis' InferenceX benchmar

Recent AI model benchmarks, such as for Alibaba's Qwen 3.8-Max and Claude Opus 5, highlight that raw scores can be misleading due to varying time and token budg

AI startup Hark, founded by Brett Adcock, has unveiled its first product, Handoff, a "computer use agent" designed to autonomously navigate the open web for use