AI Models
Frontier releases, weights, and the race to the next capability jump. ·
OpenAI Unveils GPT-6 With Native Agentic Reasoning and a 10M-Token Context Window
The new flagship plans, executes, and verifies multi-step tasks autonomously — and it runs cheaper than GPT-5.
AI ModelsOpenAI Unveils GPT-6 With Native Agentic Reasoning and a 10M-Token Context Window
The new flagship plans, executes, and verifies multi-step tasks autonomously — and it runs cheaper than GPT-5.
AI ModelsMeta Open-Sources Llama 4 Behemoth, Reigniting the Open-Weights Debate
At 2 trillion parameters, it's the largest freely available model — and a direct shot at closed labs.
AI ModelsMeta Open-Sources Llama 4 Behemoth, Reigniting the Open-Weights Debate
At 2 trillion parameters, it's the largest freely available model — and a direct shot at closed labs.
AI ModelsBenchmark Wars: Why Every Lab Now Claims State of the Art
A skeptic's guide to reading model announcements without getting fooled.
AI ModelsBenchmark Wars: Why Every Lab Now Claims State of the Art
A skeptic's guide to reading model announcements without getting fooled.
AI ModelsAnthropic’s Claude Opus 4.5 Tops the New Frontier Safety Benchmark
The model leads on refusal accuracy while cutting latency by a third.
AI ModelsQwen 3 Max Quietly Matches GPT-5 on Coding Benchmarks
Alibaba's open release narrows the East-West capability gap.
AI ModelsGoogle Ships Gemini Flash 3 for On-Device Inference
A distilled model runs fully offline on midrange phones.
AI ModelsMistral Ships a 12B Model That Beats 70B Rivals on Reasoning
Data quality, not scale, drives the latest efficiency jump.