🔍 Read the full analysis: How GPT‑6 Sol And Luna Prices Are Dropping In Half Without Affecting Performance on ThorstenMeyerAI.com
Get hardware and tech essentials delivered free — and shop member deals
- Fast, free delivery on millions of items
- Access to Prime Big Deal Days deals on October 6–7
- Prime Video, Amazon Music and more included
TL;DR
OpenAI has announced that GPT-6 Sol and Luna models are now priced at half their previous rates, with no significant decline in performance. This shift could dramatically lower AI deployment costs for businesses and developers, expanding the scope of AI applications.
OpenAI has announced a significant reduction in the prices of its GPT-6 Sol and GPT-6 Luna models, cutting costs by 50% without compromising their performance. This move aims to make advanced AI more accessible for a broader range of applications, from enterprise workflows to research. The price cuts are driven by improvements in caching and inference efficiency, according to the company, and mark a notable shift in AI affordability.
On September 22, 2026, OpenAI introduced GPT-6 Sol and Luna at half the previous prices of their GPT‑5.6 predecessors. The new pricing is $2.00 per 1 million tokens for input and $10.00 for output in Sol, and $0.10 input / $0.50 output for Luna. These reductions are achieved through enhanced caching techniques and inference improvements, which lower operational costs for OpenAI while passing savings to users.
Independent analysis by Artificial Analysis confirms that the cost per task has roughly halved, with minimal change in the models’ overall intelligence scores. GPT-6 Sol’s maximum effort score on the Artificial Analysis Intelligence Index remains high at 48, well above the median of 25, while Luna scores 37, also above its median. Despite lower costs, both models deliver comparable or improved performance in many tasks, especially in coding and hallucination reduction.
However, some regressions are noted in knowledge-based evaluations, where GPT-6 Sol and Luna scored lower on certain economic and knowledge work benchmarks. These declines are attributed to a shift in output presentation, with models producing shorter or less detailed responses, which may affect workflows requiring comprehensive deliverables. The models also exhibit increased refusal rates, declining to answer more often to reduce hallucinations, which can impact use cases involving bulk data extraction or research.
GPT‑6 Sol and Luna: half the price, about the same intelligence
OpenAI’s September 22, 2026 release doesn’t raise the ceiling. It lowers the cost of everything below it, which changes what’s worth automating.
Per 1M input / output tokens. Cached input reads keep the 90% discount.
Cost per task, halved
Measured by Artificial Analysis as the weighted cost of one Intelligence Index task, at max effort.
The effort dial moves cost more than the model choice
| Model and effort | Intelligence Index | Cost per task |
|---|---|---|
| GPT‑6 Sol (max) | 48 | $1.06 |
| GPT‑6 Sol (low) | 34 | $0.13 |
| GPT‑6 Luna (max) | 37 | $0.07 |
| GPT‑6 Luna (low) | 21 | $0.0045 |
| GPT‑6 Luna (non‑reasoning) | 18 | $0.01 |
Sol at low effort keeps about 70% of its max score for roughly an eighth of the cost, because it writes far fewer reasoning tokens. For reference, Claude Opus 5.5 leads the same index at 58.
What got better, and what got worse
Better
- Hallucination rate on AA‑Omniscience: Sol 92% → 60%, Luna 93% → 77%
- Coding Agent Index: Sol 57, up 2 points, at ~50% lower cost per task
- OpenAI reports about half as many factual mistakes for Sol as its predecessor
- Higher cache hit rates; GitHub reports over 50% fewer prompt tokens needing fresh processing
Sol gets there partly by declining more: it attempts 83% of questions vs 99%, and accuracy falls 59% → 54%.
Worse
- GDPval‑AA v2.1: Sol down ~100 Elo, Luna down ~75
- AA‑Briefcase v1.1: Luna down ~45 Elo
- Coding Agent Index: Luna 41, down 2 points
- Both models write more output tokens per task than their predecessors
Reviewers attribute the drops to weaker presentation and deliverables that omit required elements.
What to do about it
Lower Costs Expand AI Accessibility and Use Cases
The price reduction for GPT-6 Sol and Luna models represents a major shift in AI economics, lowering barriers for organizations to integrate advanced language models into their operations. By maintaining performance while halving costs, companies can now deploy AI at scale for tasks like customer support, content generation, and data analysis, previously limited by budget constraints. This development could accelerate AI adoption across industries and foster innovation in AI-powered products, though some workflows requiring detailed outputs may need adjustments due to observed regressions.
As an affiliate, we earn on qualifying purchases.
Previous Pricing and Performance Trends in GPT Models
Prior to this announcement, GPT-6 models were priced at roughly double the new rates, with GPT‑5.6 models costing $4 per 1 million input tokens and $20 per 1 million output tokens. The move to lower prices follows recent improvements in inference efficiency and caching, which OpenAI claims allow for cost-effective scaling without sacrificing quality. Historically, AI model pricing has been a key factor influencing adoption, with higher costs limiting widespread use in smaller organizations or for high-volume tasks.
OpenAI’s strategy with Astra, the top-tier model, remains focused on delivering the best possible results regardless of cost, but the new mid-tier models aim to democratize access by offering high performance at a fraction of previous prices. This aligns with broader industry trends toward making AI tools more affordable and accessible, especially as competitors like Anthropic and Google continue to innovate in model efficiency and cost reduction.
cost-effective AI development tools
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Remaining Questions on Long-Term Model Performance
It is still unclear how these models will perform over extended use cases that demand high reliability and detailed outputs, given the noted regressions in some knowledge benchmarks. The impact of increased refusal rates on workflows requiring continuous output is also yet to be fully understood. Additionally, the long-term effects of the new caching strategies on operational stability and scalability are still emerging.
As an affiliate, we earn on qualifying purchases.
Monitoring Adoption and Performance in Real-World Applications
OpenAI is expected to release further technical documentation and best practices for deploying GPT-6 Sol and Luna at scale. Industry observers and early adopters will likely evaluate how these models perform across different domains, especially in enterprise settings requiring high accuracy. Monitoring user feedback and performance metrics over the coming months will be key to understanding the full impact of these price reductions.
As an affiliate, we earn on qualifying purchases.
Key Questions
Will the performance of GPT-6 Sol and Luna decline over time?
While initial evaluations show maintained or improved performance, long-term performance in specific workflows remains to be seen, especially given some noted regressions in knowledge benchmarks.
How do the new caching improvements reduce costs?
OpenAI’s caching techniques allow for 90% discounts on cached input reads and more efficient inference, significantly lowering operational expenses and enabling price cuts.
Are these models suitable for all AI tasks?
They are optimized for many tasks, but some workflows requiring detailed, comprehensive outputs may need testing due to observed regressions in presentation quality.
Will the price cuts affect OpenAI’s top-tier Astra models?
No, Astra remains focused on delivering the highest quality results regardless of cost, with no announced changes to its pricing or capabilities.
What does this mean for AI adoption in small and medium businesses?
The significant cost reductions could make advanced AI models more accessible for smaller organizations, enabling broader integration into everyday operations.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
