Table of Contents
The biggest AI story this week is not a new capability — it is a price tag. In a stunning one-two punch, Anthropic released its flagship Claude Opus 5.5 at sharply lower rates, and within hours OpenAI answered with not one but two new GPT-6 models priced to undercut its rival. This OpenAI vs Anthropic price war marks the moment the frontier AI race officially became about cost, and creators are the biggest winners.
For anyone who uses AI tools to write, design, code, or make videos, this battle means cheaper API calls, free access to frontier models, and faster features. Here are 7 facts that explain what happened, what each lab is charging, and what it means for you.
1. Anthropic Fired the First Shot With Claude Opus 5.5
The battle began when Anthropic launched Claude Opus 5.5, its most capable model for coding and complex reasoning. Instead of raising the price for a stronger model, Anthropic cut it: Opus 5.5 costs $4 per million input tokens and $20 per million output tokens — roughly 20% below its previous rates.
That is a deliberate strategy. Anthropic is betting that cheaper premium intelligence wins more enterprise contracts, especially for coding teams and businesses running agents at scale. The company positioned Opus 5.5 as a commercial workhorse rather than a research showcase, aimed squarely at everyday business tasks.
2. OpenAI Answered Within Hours — at Exactly Half the Price
Hours after Opus 5.5 went live, OpenAI released two new models: GPT-6 Sol and GPT-6 Luna. The pricing was no accident. GPT-6 Sol costs $2 per million input tokens and $10 per million output tokens — exactly half of Opus 5.5, on both figures. GPT-6 Luna, the budget model, costs just $0.10 per million input tokens and $0.50 per million output tokens.
OpenAI priced Sol against Anthropic’s premium rival and Luna against the cheap open-weight models from China that had been competing mainly on price. In a single announcement, OpenAI moved against both ends of the market at once.
3. “Exactly Half” Is a Message, Not a Coincidence
Pricing a product at precisely 50% of a rival’s flagship on the day it launches is not subtle — it is a declaration. Industry observers read OpenAI’s move as a deliberate shot across Anthropic’s bow: whatever you charge, we can charge half.
OpenAI also cut its broader API pricing by about 50% compared with GPT-5.6 rates. The pattern is clear: the labs are no longer competing only on benchmarks and capabilities. This price war has made pricing a front-line weapon, and developers who build AI products get to pocket the difference.
4. Free ChatGPT Users Now Get Frontier AI
Perhaps the biggest win for regular creators: OpenAI opened GPT-6 to all of its roughly 1.2 billion weekly ChatGPT users, including people on the free plan. Paid subscribers on Plus, Pro, Business, and Enterprise plans can use GPT-6 Sol, while free users and subscribers on the low-cost Go plan can use GPT-6 Luna.
Until now, the newest models were reserved for paying customers. Bringing frontier AI to free users changes the math for anyone making content on a budget — you no longer need a subscription to use OpenAI’s latest model.
5. The New Models Come With an “Intelligent UI”
Price was not the only news. The GPT-6 update brings what OpenAI calls an “Intelligent UI” — a major change in how ChatGPT answers. Instead of text-only replies, ChatGPT can respond with charts, diagrams, buttons, and forms depending on the question.
Ask about comparisons and it shows information side by side. Plan a trip and it organizes routes and destinations on a map. For cooking, it lays out the steps and timing together. ChatGPT can even create small interactive tools on demand, like a savings calculator, a cost-splitting calculator, or a simple game.
We covered this interface change in detail in our guide to GPT-6 Intelligent UI, which breaks down exactly how the new screens work.
6. Anthropic’s Budget Move: Claude Haiku 5.5
Anthropic did not only fight at the premium end. On the same day, it launched Claude Haiku 5.5, a small model with running costs about 75% lower than its predecessor. Haiku 5.5 costs just $0.10 per million input tokens on prompts up to 100K tokens — 90% below Haiku 4.5.
Anthropic says it is the fastest model it has ever released, with adjustable effort settings from Low to Max so teams can tune cost against intelligence per task. Early customers report real gains: Asana saw over 30% lower task-completion latency in its AI Teammates evaluations, and HubSpot recorded 92.8% on its CRM test suite — the best it has seen from a smaller model.
If you want the full rundown on the small model, read our Claude Haiku 5.5 breakdown.
7. What This Price War Means for Creators
So who wins this OpenAI vs Anthropic price war? Short answer: you do. Cheaper frontier models mean lower costs for anyone running AI-powered apps, automations, or content pipelines. A creator generating thousands of images, scripts, or video captions a month can now cut the AI bill dramatically — or get far more output for the same budget.
It also means more competition. With Google, Meta, OpenAI, and Anthropic all racing to win developers, expect faster releases, better free tiers, and new features bundled into the same subscription you already pay for. The labs are fighting for your loyalty now, and loyalty gets cheaper every quarter.
There is a practical side to this too. If you run a small business or side project powered by AI — a chatbot, a content automation, a research assistant — now is a good time to compare plans. Switching an agent workflow from a premium model to GPT-6 Luna or Claude Haiku 5.5 for routine tasks can cut inference costs to a fraction of what they were last year, while reserving the bigger models for the hard problems. Many teams now mix and match: cheap models for bulk work, flagship models for judgment calls.
FAQ
Why did OpenAI price GPT-6 Sol at exactly half of Claude Opus 5.5?
OpenAI launched GPT-6 Sol at $2 per million input tokens and $10 per million output tokens, exactly half of Anthropic’s Claude Opus 5.5 pricing ($4 input, $20 output). The timing — hours after Opus 5.5 launched — shows it was a deliberate undercut aimed at winning developers.
Can free ChatGPT users access GPT-6?
Yes. OpenAI made GPT-6 available to all roughly 1.2 billion weekly ChatGPT users. Free users and Go-plan subscribers can use GPT-6 Luna, while paid Plus, Pro, Business, and Enterprise subscribers can use the more capable GPT-6 Sol.
What is the cheapest frontier AI model right now?
Among the new launches, GPT-6 Luna and Claude Haiku 5.5 both start at $0.10 per million input tokens, undercutting cheap open-weight models. For premium coding and reasoning, GPT-6 Sol at $2/$10 currently beats Opus 5.5 at $4/$20 on price.
Is the AI price war good for creators?
Yes. Lower model prices cut the cost of running AI apps, automations, and content workflows. Combined with free access to frontier models and new features like the Intelligent UI, creators get more capability for less money — with more competition still to come.
The frontier model race used to be a contest of capability. The OpenAI vs Anthropic price war shows it is now a contest of value too. To see how the models at the center of this battle stack up feature by feature, check our GPT-6 Sol and Luna guide — and keep an eye on prices, because this fight is just getting started.
For official product details and pricing, visit OpenAI.