🔍 Read the full analysis: OpenAI Cuts GPT‑6 Sol And Luna Prices In Half, Benchmark Results Stay Steady on ThorstenMeyerAI.com
Get audio and creator gear 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 cut the prices of its GPT‑6 Sol and Luna models by 50%, while benchmark scores remain largely unchanged. The move aims to make AI more affordable without sacrificing performance, potentially broadening adoption across industries.
OpenAI has slashed the prices of its GPT‑6 Sol and Luna models by 50%, effective immediately, while maintaining comparable benchmark performance. This move aims to significantly lower the cost barrier for deploying advanced AI, making it accessible to a broader range of businesses and applications. The price reduction was announced on September 22, 2026, and is framed as a strategic effort to democratize AI benefits through improved cost efficiency.
OpenAI’s new pricing structure reduces the cost per 1 million tokens for GPT‑6 Sol from $4 to $2 for input and from $20 to $10 for output. Similarly, GPT‑6 Luna’s prices have been halved from $0.20 to $0.10 for input and from $1.20 to $0.50 for output. The reductions are attributed to improvements in caching and inference technologies, which allow OpenAI to serve these models at lower costs, passing the savings directly to customers.
Independent analysis by Artificial Analysis confirms that while costs per task have roughly halved, the models’ performance on benchmark tests remains steady or slightly improved in some areas. For example, GPT‑6 Sol scores 48 on the Artificial Analysis Intelligence Index, well above the median of 25 for comparable models, with Luna scoring 37 against a median of 12. These scores indicate maintained or improved capabilities despite the lower prices.
However, some evaluations reveal regressions, particularly in knowledge-based tasks, where GPT‑6 models showed decreased performance in economic and knowledge work benchmarks. OpenAI attributes these regressions to changes in presentation quality and output length, which may impact workflows requiring detailed, well-structured deliverables.
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
Implications for AI Deployment and Cost Efficiency
The price reductions for GPT‑6 Sol and Luna could dramatically lower the cost barrier for integrating advanced AI into products, services, and workflows. This shift may accelerate AI adoption across industries, especially for applications where cost constraints previously limited use. It also signals OpenAI’s focus on cost efficiency as a key strategic goal, emphasizing broader distribution of AI benefits without sacrificing model performance.
Businesses that rely on large-scale AI tasks, such as customer support, content generation, or data analysis, may find it more feasible to scale their AI operations. The move could also intensify competitive pressures among AI providers, prompting others to adjust their pricing strategies. Nonetheless, the sustained benchmark performance suggests that cost savings do not come at the expense of core capabilities, making these models attractive options for a wide range of applications.
As an affiliate, we earn on qualifying purchases.
Background on GPT‑6 Pricing and Performance
OpenAI introduced GPT‑6 Astra earlier in 2026, marking a new high-performance model with a focus on intelligence and capacity. The GPT‑6 Sol and Luna models, launched on September 22, 2026, are positioned as more affordable options within the same family, emphasizing cost efficiency over pushing the upper limits of capabilities. Prior to this, GPT‑5.6 models were priced higher, with the new models now offering roughly 50% lower costs.
The models’ performance has been evaluated through independent benchmarks, including Artificial Analysis’s Intelligence Index and various coding and knowledge-work tests. While some metrics show stable or improved performance, certain areas, notably detailed knowledge tasks, have experienced regressions, likely due to tuning aimed at reducing verbosity and improving user experience.
OpenAI’s improvements in caching and inference technology have been central to enabling the price cuts, with features such as 90% discounts on cached input reads and new diagnostics tools aiding developers in optimizing usage and costs.
Uncertainties About Long-term Performance and Adoption
It is still unclear how the reduced prices will affect long-term performance in real-world applications, especially in complex knowledge tasks where some regressions have been observed. The impact on adoption rates across different industries remains to be seen, as organizations may need time to validate the models in their workflows. Additionally, the durability of the performance benchmarks over time and across diverse use cases is yet to be confirmed.
Next Steps for OpenAI and Users
OpenAI is expected to continue refining its models and infrastructure, possibly releasing further updates to optimize performance and cost. Users and organizations should conduct thorough testing of the new models within their specific workflows before full adoption, especially for tasks requiring detailed outputs. Monitoring performance over the coming months will be crucial to understanding the full impact of these price cuts on operational efficiency and AI integration strategies.
Key Questions
Will the performance of GPT‑6 Sol and Luna decline over time?
While current benchmark results are stable or improved, long-term performance in specific real-world tasks remains to be seen. Ongoing updates and testing will clarify this in the coming months.
How will the price cut affect AI adoption in small and medium-sized businesses?
The reduced costs could make advanced AI models more accessible to smaller organizations, enabling broader deployment across various sectors, from customer service to content creation.
Are there any trade-offs in quality or capabilities due to the price reduction?
Some evaluations indicate minor regressions in certain knowledge tasks, likely due to tuning for better user experience. Overall, core capabilities appear maintained, but users should test models for their specific needs.
Will OpenAI continue lowering prices for other models?
OpenAI has not announced further price cuts beyond GPT‑6 Sol and Luna at this time, but the focus on efficiency suggests potential future adjustments.
What should organizations do before fully adopting the new models?
Organizations should conduct pilot tests, evaluate performance in their specific workflows, and monitor outputs closely to ensure the models meet their quality and reliability standards.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
