OpenAI brought its newest model to everyone this week. GPT-6 arrived in ChatGPT on October 7 with a feature called Intelligent UI: instead of replying only in text, ChatGPT now picks a format for each question, from charts and forms to tappable buttons, and it can start answering while it keeps thinking. On web-search questions it begins answering 44% sooner than the previous version. Paid plans got it October 7, free plans October 8, reaching all 1.2 billion weekly ChatGPT users.
The cheap end of the market fell further. Anthropic’s Claude Haiku 5.5 costs about 75% less to run than Haiku 4.5 and scores 1,620 on a knowledge-work test called GDPval-AA v2.1, up from 735. Mistral previewed Mistral Large 4, a trillion-parameter model with 52 billion active, trained on 3,800 Nvidia chips in Europe, weights promised by the end of October. Reflection released Beam, its first open-weight model at 501 billion parameters, and five days later the Financial Times reported Nvidia is in talks to buy Reflection or deepen its stake.
The push to slow AI down also got louder. Team Human, a campaign backed by creators with more than 300 million subscribers, is collecting names to deliver to lawmakers in Washington and Brussels, asking governments to agree on a global slowdown and to track the chips that train frontier AI as nuclear materials are tracked. Three safety researchers OpenAI had fired published an open letter denying misconduct, warning the dismissals would chill safety work inside the company. Capability keeps getting cheaper and more available; who decides how fast to push it is still open.
Read More: AI This Week: Cheaper Flagships, Always-On Agents, and a Fight Over Open Weights
Sources:
- GPT-6 and Intelligent UI for everyone (OpenAI)
- Claude Haiku 5.5 (Anthropic)
- Introducing Mistral Large 4 (Mistral)
- Introducing Beam: Reflection’s 501B open-weight model (Reflection)
- Nvidia in talks to invest further in Reflection AI or buy it, FT reports (Reuters)
- Team Human: Keep Humans in Control of AI
- Fired OpenAI safety researchers dispute misconduct claims, warn of chilling effect (TechCrunch)
Disclaimer: For information only. Accuracy or completeness not guaranteed. Illegal use prohibited. Not professional advice or solicitation. Read more: /terms-of-service
Reuse
Citation
@misc{kabui2026,
author = {{Kabui, Charles}},
title = {AI {This} {Week:} {GPT-6} for {Everyone,} a {Cheaper}
{Floor,} and a {Push} to {Slow} {Down}},
date = {2026-10-11},
url = {https://toknow.ai/posts/ai-weekly-gpt6-for-everyone-haiku-5-5-slowdown/},
langid = {en-GB}
}
