AI Price War Heats Up: OpenAI and Anthropic Cut Model Prices on the Same Day

OpenAI and Anthropic Cut Prices — On the Exact Same Day

Something unusual happened this past Tuesday (Sept 22): Anthropic and OpenAI, the two leading AI labs, both announced cheaper new models within less than 90 minutes of each other.

Anthropic released Claude Opus 5.5 — its most capable model — at a 40% lower price than its predecessor ($20 per million output tokens vs. $25 for Opus 5), faster to run, and according to the company outperforming its larger Fable model on several benchmarks. This is Anthropic's first release since CEO Dario Amodei publicly committed to "pacing the frontier" — deliberately slowing capability advances to let safety work keep up.

About 90 minutes later, OpenAI launched GPT-6 Sol and GPT-6 Luna — updated versions of its mid-tier and lightweight models — at half the price of the previous (5.6) generation, with roughly half the factual error rate. OpenAI attributes the price drop mainly to improvements in caching and inference efficiency.

Why this matters for small and medium businesses: this price competition directly benefits anyone using AI for everyday work — writing, document summarization, coding, customer support. Cheaper models with fewer factual errors lower the cost of adoption and reduce the risk of integrating AI into business workflows. It's worth revisiting your current AI spend against actual needs — a new "smaller" model may now handle most day-to-day tasks at a significantly lower cost.

Sources: TechCrunch — Anthropic, TechCrunch — OpenAI

Meta Announces: Personal Agent Muse Comes to Glasses, Mac, and Small Businesses

At Meta Connect 2026 (Sept 23), Mark Zuckerberg laid out Meta's full plan for its personal AI agent, Muse: a realtime video-chat avatar, full integration with Meta's smart glasses (wake-word activation, background task handling), the ability to autonomously run tasks on a Mac, and its own dedicated email address. A Meta executive stated explicitly: "you can have it help you run your small business." Meta also locked in partnerships with Shopify, Stripe, PayPal, and retailers like Best Buy, Sephora, and Walmart, and opened a developer connector platform that received over 1,500 applications in its first week.

Why it matters: another sign that personal AI agents are shifting from "chatbot" to a digital assistant that actually handles tasks, payments, and communication — a trend directly relevant to small businesses looking for accessible automation.

Source: TechCrunch

Anthropic's Wet Lab Reports Its First Discovery

In the physical biology lab Anthropic confirmed operating in the Bay Area last week, the company reports its first finding: a previously unknown enzyme system in bacteriophage DNA (viruses that infect bacteria) with properties "reminiscent of CRISPR" — the ability to cut, copy, and paste DNA. According to Anthropic, the discovery process was driven mostly by Claude (about 950 agents, 210 million tokens, 21 hours of work), though the actual physical lab experiments were carried out by human scientists only. Anthropic CEO Dario Amodei notes the finding is similar to a previous discovery by a Stanford research group, and that its scientific significance still needs validation by the broader research community.

Source: TechCrunch

In Brief

• 20 countries (excluding the US and China), including Canada, Germany, and the UAE, signed onto a call for stronger oversight of advanced AI models.
• Senator Bernie Sanders introduced legislation to ban development of "superintelligence" with criminal penalties — another sign of the escalating US political debate on AI safety.