Editorial illustration for OpenAI's GPT-6 Models Halve Prices but Show Minimal Performance Gains
OpenAI's GPT-6 Models Cut Prices in Half
OpenAI's GPT-6 Models Halve Prices but Show Minimal Performance Gains
OpenAI cut prices in half on two new GPT-6 models without moving performance much beyond what GPT-5.6 already delivered. Sol and Luna, announced this week, replace Terra as the budget end of the GPT-6 lineup, with Sol priced at $2 per million input tokens and $10 per million output tokens, and Luna at $0.10 and $0.50 respectively. Both figures are half of what OpenAI charged for the equivalent GPT-5.6 models.
The company says the drop comes from gains in caching and inference rather than any shortcut on quality, and that it's passing those savings straight to developers. Sol is pitched at recurring, complex work: feature development, code review, debugging, data analysis. Luna is built for high-volume, low-stakes tasks like summarizing text or pulling structured information out of documents. Pricing now puts OpenAI closer to what cheaper open-weight models charge, and within range of some of Anthropic's costlier offerings.
Both models launch inside ChatGPT Work and Codex, available to Plus, Pro, Business, Enterprise, and Edu subscribers. The bigger question is what OpenAI actually changed under the hood to make that price cut possible.
According to Artificial Analysis, GPT-6 Sol and Luna cut per-task costs in half compared to their predecessors, but intelligence scores stay at GPT-5.6 levels, with gains in some evaluations and regressions in others.
Why this matters
For developers and founders running production workloads, GPT-6 Sol and Luna are a signal that OpenAI has stopped competing purely on raw capability and started competing on margin. Cutting Sol to $2/$10 per million tokens and Luna to $0.10/$0.50 while holding benchmarks roughly flat tells us the model itself isn't the story here, the unit economics are. If your app calls GPT-5.6 today, this is a straightforward cost cut with no migration risk worth worrying about.
The OSWorld 2.0 comparison against Claude Opus 5 is worth watching closely, though. An 80 percent cost advantage at similar computer-use performance is a real pricing weapon against Anthropic, and it will pressure Anthropic's own pricing or force them to justify Opus 5's premium on capability alone. For researchers, the flat benchmark curve across a full price generation is the more interesting data point: it suggests the low-hanging gains in this model tier are gone, and OpenAI is now optimizing serving cost rather than raw performance. Watch how Anthropic responds on price before assuming this shifts market share.
Related Reading
Browse all OpenAI news →
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google tests visual 'magazine-style' UI for Gemini 3 Pro users
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Related Topic OpenAI's GPT Transcribe Cuts Error Rate to 3.31%, Improving on GPT-4o
Open Source
Related Topic OpenAI Developer: AI Agent "Swarms" Waste Tokens, Create "Coordination Tax
AI Tools & Apps
From Archives Claude Opus 5.5 Cuts Costs By Using 40% Fewer Tokens
From Archives Anthropic Claims Claude Opus 5.5 Beats GPT-6 Astra at 20% of the Cost
From Archives The Download: AI's self-improvement problem and OpenAI's safety pause
Research & Benchmarks
From Archives OpenAI Claims Its Model Solved Over 100 Math Problems in a Month
Research & Benchmarks
Common Questions Answered
What are the pricing differences between GPT-6 Sol and Luna compared to GPT-5.6?
GPT-6 Sol is priced at $2 per million input tokens and $10 per million output tokens, while Luna costs $0.10 and $0.50 respectively. Both models represent a 50% price reduction compared to their GPT-5.6 equivalents, making them significantly more affordable options for developers running production workloads.
How do GPT-6 Sol and Luna perform compared to GPT-5.6 on intelligence benchmarks?
According to Artificial Analysis, GPT-6 Sol and Luna maintain intelligence scores at GPT-5.6 levels with minimal performance gains. The models show gains in some evaluations and regressions in others, indicating that OpenAI prioritized cost efficiency over raw capability improvements in this release.
What drove OpenAI's decision to cut prices in half for the new GPT-6 models?
OpenAI attributed the price reduction to gains in caching and inference optimization rather than improvements in the underlying model performance. This shift suggests the company is now competing on unit economics and margin rather than purely on raw capability advancement.
Should developers currently using GPT-5.6 migrate to GPT-6 Sol or Luna?
For developers running production workloads with GPT-5.6, migrating to GPT-6 Sol or Luna represents a straightforward cost reduction with no migration risk worth worrying about. Since the models maintain similar performance levels, the primary benefit is achieving significant savings on per-task costs without sacrificing functionality.
Further Reading
- GPT-5.6: Frontier intelligence that scales with your ambition - OpenAI
- Advancing the price-performance frontier with GPT-5.6 - OpenAI
- OpenAI cuts prices for two of its GPT-5.6 AI models as companies grow sensitive to costs - CNBC
- AI price wars: OpenAI cuts GPT-5.6 Luna prices by 80% as model competition shifts toward cost - VentureBeat
- OpenAI cuts GPT-5.6 Luna and Terra prices by up to 80% - Yahoo Finance
