ResearchAudio

Evidence-led AI briefings and free decision tools for engineers, researchers, and product builders.

Latest

DeepSeek's New Base Loses 12 of 16 Rows to the Model It Replaces

Sep 11, 2026

DeepSeek claimed to beat Anthropic and OpenAI

Tencent's World Generator Runs on Claude Opus 4.8

Sep 9, 2026

GPT-Image-2 paints objects onto a render, SAM3D lifts them. The evaluation has no numbers.

Astra and Fable 5.1 Share a Rate Card, Not a Bill

Sep 7, 2026

Same 10 in and 50 out per million tokens. One burns 3.3x the tokens per index task and bills 2.4x. One cache line sits 4x apart.

Why AI agents keep relearning the same lessons

Sep 5, 2026

A study turns software knowledge into tested instructions for research agents. The gains are substantial. Understanding where they come from takes a closer look.

OpenAI's Most Aligned Model Also Writes the Least Down

Sep 4, 2026

Scope violations fell to 0%. Reasoning monitorability fell too. Your agent can get a 403.