Mistral AI introduced two new transcription models—Voxtral Mini Transcribe 2 and Voxtral Realtime—designed to run on edge devices such as phones, laptops, and wearables. The compact models prioritize privacy by keeping data local, and they deliver low‑latency performance, with the realtime model achieving less than 200 milliseconds of delay. Available via Mistral’s API and on Hugging Face, the models support 13 languages and can be customized for specific vocabularies, offering accuracy comparable to larger systems while maintaining speed and user control. Leia mais →

OpenAI announced the release of GPT-5.2, its most advanced model to date, as part of a company‑wide “code red” effort to strengthen ChatGPT against mounting competition from rivals like Google. The new series includes Instant, Thinking and Pro tiers, each tuned for speed, coding and high‑accuracy tasks. GPT-5.2 shows notable gains on the GDPval benchmark, beating human professionals on a majority of tasks, and reduces hallucinations by a reported 38 percent. OpenAI also highlighted new safeguards for mental‑health‑sensitive prompts and plans for an “adult mode” later next year. Leia mais →

OpenAI has introduced two updated versions of its flagship model, GPT-5.1 Instant and GPT-5.1 Thinking, now available in ChatGPT. The rollout adds eight preset personality styles—Professional, Friendly, Candid, Quirky, Efficient, Cynical, Nerdy, and Default—and incorporates adaptive reasoning that lets the model allocate extra compute time when needed. OpenAI says the new models outperform previous versions on technical benchmarks such as AIME and Codeforces. The release follows user complaints about earlier model tone and comes amid heightened regulatory scrutiny. Paid subscribers receive access first, with broader availability and API integration planned shortly. Leia mais →