ChatGPT Got Better and Cheaper This Week. The Price War Continues.

Two weeks ago, I wrote a blog post on Galyx about the AI price war that was kicking off between Anthropic’s Fable 5 and Moonshot AI’s Kimi K3. The argument was simple: when frontier models compete on price, the entire market gets cheaper. Business owners benefit, whether they use the premium tier or not.

I said the pricing collision would determine what your tools cost for the next 12 months. This week, OpenAI proved the point.

On August 6, OpenAI announced a set of changes to ChatGPT that affect every tier: free, paid, and developer. The improvements are real, the pricing shifts are significant, and the strategic positioning tells you a lot about where this market is headed. If you use ChatGPT for any part of your business, the decision about what you pay and what you get back changed this week.

What OpenAI Changed on August 6

Three things happened at once; let’s look at them separately.

First, OpenAI tuned GPT-5.6 Sol, their flagship model, specifically for everyday ChatGPT conversations. Sol was already the most capable model in the GPT-5.6 family, but this update optimized it for the tasks most people use ChatGPT for: quick questions, web research, planning, writing, and advice. The tuned version delivers more focused answers, adapts its level of detail to the question, reduces unnecessary formatting, and, according to OpenAI’s own internal evaluation, produces factual errors 68% less often than the previous GPT-5.5 Instant on high-stakes prompts in finance, medicine, and law.

Second, free and Go-tier users now get GPT-5.6 Luna as their default model, replacing whatever they had before. Luna is the most affordable model in the GPT-5.6 family, but it’s still a meaningful upgrade from GPT-5.5. OpenAI claims Luna produces factual errors 62% less often than GPT-5.5 Instant on the same high-stakes prompt set. Free users also get a new “Think” button that lets Luna take more time on harder questions, giving it access to deeper reasoning when you need it.

Third, and this is the headline for anyone who has bumped into ChatGPT’s rate limits: free-tier text conversations are becoming unlimited. No more hitting a wall after a handful of messages and being told to come back later. OpenAI is rolling this out over the next week, subject to what they call “abuse guardrails,” but the directional change is clear. The free tier is no longer a demo. It’s a powerful product.