MarkTechPost (1 month ago)
Most teams treat ‘which model’ as the important decision. The harness engineering literature keeps pointing somewhere else. In LangChain’s Terminal-Bench experiment, changing only the harness—same model throughout—moved a coding agent from roughly 30th place into the top 5.
Read the full story on MarkTechPost.
The FBI arrested a Saudi national who worked at OpenAI on the product team. The AI maker accused the employee of hiding a recording device in an executive conference room.
Judge Edward Davila today granted some of Apple’s requests for expedited discovery in its trade secrets lawsuit against former employees, OpenAI, and io Products. Here are the details. more…
An OpenAI engineer was arrested after the company said it discovered a “sophisticated recording device” hidden in one of its offices.
A federal judge ruled that Apple can access OpenAI’s interview materials involving former Apple employees, as well as private messages and testimony from key figures in the case.
Its prices make it competitive for developers and businesses willing to pay for their AI use, but the lack of a free option will likely hamstring mainstream appeal.
Mark Zuckerberg says Meta committed as much as $200B to AI infrastructure before anyone could tell him it would work, with his chief AI officer saying it was not clear at all. Meta’s free cash flow fell 91% in the second quarter.
Z.ai’s GLM-Legal is a legal AI model built on GLM-5.3-Flash. It scores 19.2% all-pass on Harvey’s Legal Agent Benchmark, nearly 3x its base, at $0.15 per test. Access is by application.
Alibaba’s Qwen team has introduced QwenGyre, an elastic reinforcement learning RL framework for extreme-long-horizon agent training. QwenGyre is an online reinforcement learning system from Alibaba that trains coding agents through unmodified, black-box harnesses.
We checked 12 AI live translation tools, including Palabra.ai, Google Live Translate, DeepL Voice, Teams Interpreter and Wordly, against official pages for languages, latency, platforms and pricing.
TII has released Falcon-ASR, a 1.6B speech recognition model for Arabic with an Emirati focus. It reports 20.92% average WER on 6 Arabic test sets, but weights and an API are not yet available.
72 active sources
27 Turkish · 45 English
All sources
