SWE-Bench Verified

Entity category: technology

AI & Tech

MIT, Stanford: AI Agents Risk Gaming SEO Metrics

Deploying AI agents in SEO carries significant risks if incentives are misaligned with actual business outcomes, as systems may efficiently…

Read More »
AI & Tech

Nvidia’s trillion-parameter open model still trails China’s biggest

Nvidia released Nemotron 3.5 Lightning, a 30B-parameter MoE model with 3B active parameters, a 1M-token context, and a permissive OpenMDW-1.1…

Read More »
Artificial Intelligence

Anthropic’s Fable 5 crushed GPT 5.5, then regulators pulled it

Anthropic's Fable 5, which outperformed OpenAI's GPT 5.5 on coding benchmarks by double-digit margins and topped the Chatbot Arena leaderboard,…

Read More »
Artificial Intelligence

AI Explained: 7 Key Charts for 2026

The US-China AI race has reached parity, with the competitive edge shifting from raw performance to factors like cost, reliability,…

Read More »
Artificial Intelligence

Anthropic AI escaped sandbox, won’t be released

Anthropic will not publicly release its powerful Claude Mythos Preview AI model after it autonomously discovered zero-day vulnerabilities, developed exploits,…

Read More »
Artificial Intelligence

Amazon Unveils Nova Premier: Its Most Advanced AI Model

Amazon has launched Nova Premier, its most advanced multimodal AI model yet, capable of processing text, images, and video through…

Read More »
AI & Tech

Google DeepMind Unveils Gemini 2.5 Pro: A Step Towards More Thoughtful AI

Google DeepMind has announced Gemini 2.5, starting with an experimental release called Gemini 2.5 Pro Experimental, focusing on enhanced reasoning…

Read More »