🔍 Read the full analysis: The Price Drop Of GPT‑6 Sol And Luna: What It Means For AI Developers 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 announced a 50% price reduction for GPT‑6 Sol and Luna models, aiming to expand AI deployment by making models more affordable. The models show comparable performance with improved hallucination management, though some regressions in knowledge tasks are noted. The move could significantly lower AI adoption barriers for developers.
OpenAI has announced a significant price cut for its GPT‑6 Sol and GPT‑6 Luna models, reducing costs by approximately 50% compared to previous GPT‑5.6 versions. The models, released on September 22, 2026, are designed to make advanced AI more accessible for a broader range of developers and organizations, emphasizing cost efficiency over new capabilities. This move aims to accelerate AI adoption by lowering financial barriers for deploying large language models in various applications.
Both GPT‑6 Sol and Luna now cost half their previous prices, with Sol priced at $2.00 per 1 million input tokens and $10.00 per 1 million output tokens, and Luna at $0.10 and $0.50 respectively. The price reductions are achieved through improvements in caching and inference techniques, which reduce operational costs for OpenAI, with savings passed directly to customers.
Independent analysis by Artificial Analysis confirms that the cost per task has halved, with GPT‑6 Sol at maximum effort costing about $1.06 per task, down from roughly $1.99 for GPT‑5.6 Sol, and Luna at about $0.07 per task, a 60% reduction. Performance metrics show that both models maintain strong scores on intelligence benchmarks, with Sol scoring 48 and Luna 37 on the Artificial Analysis Intelligence Index, significantly above median scores for their price classes.
In terms of quality, both models exhibit improved hallucination rates, reducing incorrect or fabricated answers. Sol’s hallucination rate dropped from 92% to 60%, and Luna’s from 93% to 77%. However, these improvements come with trade-offs; Sol now declines to answer more questions, reducing its answer rate from 99% to 83%, which lowers hallucinations but also slightly decreases accuracy. Some regressions in knowledge-task performance have been observed, attributed to changes in output presentation quality, especially for complex, detailed responses.
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
Impact of Price Reduction on AI Deployment Strategies
The 50% price cut for GPT‑6 Sol and Luna models is a pivotal development for AI developers and businesses. Lower costs make it feasible to deploy large language models at scale in customer service, automation, research, and content generation, broadening AI’s reach beyond high-budget enterprises. The improved cost-efficiency could accelerate innovation, enable new use cases, and democratize access to advanced AI capabilities. However, the slight regressions in knowledge accuracy and output quality for complex tasks suggest that users should carefully evaluate whether these models meet their specific needs, especially for tasks requiring high precision or detailed outputs.
This shift could also influence the competitive landscape, prompting other AI providers to adjust their pricing strategies. For organizations already using GPT‑6 models, the reduced costs may allow for larger or more frequent deployments, potentially transforming operational workflows and product offerings. The move underscores a broader industry trend toward making AI more affordable and accessible, which could accelerate adoption across sectors.
As an affiliate, we earn on qualifying purchases.
Background on GPT‑6 Price and Performance Evolution
OpenAI introduced GPT‑6 Astra earlier in September 2026, which set a new benchmark for model intelligence and capability. The subsequent release of GPT‑6 Sol and Luna focused primarily on cost reduction, not capability enhancements. Prior to this, GPT‑5.6 models were priced higher, limiting their use to organizations with substantial budgets. The improvements in caching and inference efficiency that enable the price cuts were developed over recent months, reflecting ongoing efforts to optimize operational costs.
Independent evaluations, such as those from Artificial Analysis, have consistently shown that while the newer models maintain or slightly improve performance on many benchmarks, some knowledge-based tasks have experienced regressions. These changes are attributed to tuning aimed at reducing verbosity and enhancing user experience, which sometimes impacts detailed output quality. The industry has seen a broader trend of balancing performance with cost and usability, with OpenAI leading efforts to make high-quality AI more affordable.
Overall, this development marks a shift toward more accessible AI, aligning with OpenAI’s mission to democratize artificial intelligence and foster widespread adoption.
As an affiliate, we earn on qualifying purchases.
Unresolved Questions About Long-term Performance
It remains unclear how sustained these performance levels will be across diverse real-world applications, especially for complex knowledge tasks. The observed regressions in some benchmarks suggest that further tuning may be needed to balance cost, accuracy, and output quality. Additionally, the long-term impact on competitive positioning and whether other providers will follow suit with similar price cuts are still developing stories.
Furthermore, the extent to which these models can replace more specialized or larger models in critical applications remains uncertain, as some trade-offs in detailed output quality and knowledge accuracy are still evident.
AI model caching and inference hardware
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps for AI Developers and OpenAI
Developers should evaluate these models within their specific workflows, especially testing for detailed, knowledge-intensive tasks to ensure performance meets requirements. OpenAI is expected to continue optimizing caching and inference techniques, potentially further reducing costs and improving performance. Monitoring user feedback and independent evaluations will be crucial to gauge the models’ practical effectiveness over time.
OpenAI may also release updates or new models that address current regressions, and industry competition could lead to further price adjustments. The broader AI ecosystem is likely to see increased adoption, with more organizations integrating these cost-effective models into their products and services.
affordable AI chatbot development kits
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Key Questions
How much cheaper are GPT‑6 Sol and Luna compared to previous models?
GPT‑6 Sol’s input cost is $2.00 per 1 million tokens, down from $4, and output is $10.00, down from $20. GPT‑6 Luna’s input cost is $0.10, down from $0.20, and output is $0.50, down from $1.20, representing about a 50% to 60% reduction.
Do the new models perform as well as previous versions?
Performance on many benchmarks remains comparable or better, with notable improvements in hallucination reduction. However, some knowledge-based tasks have seen regressions, especially in detailed output quality, which users should evaluate for their specific needs.
What are the main benefits of the price reduction for developers?
The lower costs enable larger-scale deployment, more frequent usage, and broader application of GPT‑6 models, making advanced AI more accessible for startups, SMEs, and large enterprises alike.
Are there any downsides to the new models?
Some regressions in knowledge accuracy and output detail have been observed, and models now decline to answer more often, which might impact workflows requiring comprehensive responses. Careful testing is recommended before full adoption.
What is likely to happen next in this development?
OpenAI will probably continue refining these models and their caching techniques, possibly releasing further updates. Industry competition may lead to additional price cuts, expanding AI adoption even further.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
