🔍 Read the full analysis: Half-Price GPT‑6 Sol And Luna: The Benchmark Effect Of OpenAI’s Pricing Strategy on ThorstenMeyerAI.com
TL;DR
OpenAI introduced GPT-6 Sol and Luna models at half the previous pricing, emphasizing cost efficiency without sacrificing performance. This shift could significantly influence AI deployment strategies and automation affordability.
OpenAI has released new versions of its GPT-6 models, GPT-6 Sol and GPT-6 Luna, at approximately half their previous prices. The move aims to make AI more affordable for a wider range of applications, with the models now available at significantly lower costs, while maintaining comparable performance levels. This development is expected to influence how businesses and developers incorporate AI into their workflows and products, potentially expanding automation and AI-driven solutions across various industries.
Both GPT-6 Sol and Luna arrived on September 22, 2026, with prices reduced by 50% compared to GPT‑5.6 models. The pricing adjustments are primarily driven by improvements in caching and inference efficiency, enabling OpenAI to lower operational costs and pass these savings onto customers. GPT-6 Sol now costs $2.00 per 1 million tokens for input and $10.00 for output, while GPT-6 Luna costs $0.10 for input and $0.50 for output, representing a significant reduction from previous models.
Independent analysis by Artificial Analysis confirms that the cost per task has halved, with consistent performance scores. GPT‑6 Sol scores 48 on the Artificial Analysis Intelligence Index, well above the median of 25, while Luna scores 37, compared to a median of 12. Despite the lower prices, the models demonstrate only minor regressions in some knowledge-based evaluations, attributed to changes in output presentation rather than core capabilities.
In coding tasks, Sol scores 57 on the Coding Agent Index, slightly above its predecessor, with improvements in some benchmarks. Hallucination rates have decreased notably, with Sol reducing errors from 92% to 60%, and Luna from 93% to 77%, indicating enhanced factual accuracy. However, some performance regressions in knowledge work and output quality have been observed, particularly in detailed deliverables, which may affect workflows requiring comprehensive outputs.
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 Cost and Deployment Strategies
The release of GPT-6 Sol and Luna at half-price marks a pivotal shift in AI economics, making advanced language models more accessible for a broad array of applications. By significantly reducing costs, businesses can now automate tasks previously deemed too expensive, potentially expanding the scope of AI integration in customer service, research, and content generation. This move could accelerate AI adoption, lower barriers for startups and smaller enterprises, and influence competitors to adjust their pricing strategies, ultimately shaping the AI market landscape.
Furthermore, the models’ comparable performance at lower costs challenges assumptions about the necessity of high-end, expensive AI for effective automation. It underscores the importance of cost-efficiency innovations—like caching and inference improvements—in democratizing AI technology. However, the observed regressions in some knowledge tasks suggest that users should evaluate these models carefully for specific workflows, especially those requiring detailed, high-quality outputs.
Popular right now
KitchenAid Classic Series 4.5-Quart Tilt-Head Stand Mixer, Onyx Black
Dyson R3 Nurovi Spot+Scrub UV Robot Vacuum
iRobot Roomba Plus 678 Robot Vacuum and Mop with AutoWash Dock
As an affiliate, we earn on qualifying purchases.
OpenAI’s Pricing Strategy and Model Development Timeline
OpenAI’s move to halve the prices of GPT-6 Sol and Luna follows a series of strategic developments aimed at broadening AI accessibility. Previously, GPT‑5.6 models served as the benchmark for high-performance language models, with pricing reflecting their advanced capabilities. The release of Astra, a more powerful model announced earlier in 2026, set a new performance standard, but the real breakthrough was the focus on cost reduction for mid-tier models.
OpenAI attributes the price cuts to advances in caching, inference, and system optimization, which have allowed the models to operate at lower costs without sacrificing core performance. This strategy aligns with broader industry trends emphasizing democratization of AI, where affordability becomes a key driver of adoption. The timing coincides with increased competition from other AI providers, such as Anthropic’s Claude Opus 5.5, which also announced price cuts, indicating a market-wide shift towards more accessible AI tools.
These developments reflect OpenAI’s dual focus: maintaining top-tier models like Astra for high-stakes applications while pushing affordability in the broader market segment. The models’ release also follows a period of extensive internal testing and independent evaluations, which confirm that the models deliver similar capabilities at a fraction of previous costs.
Unanswered Questions About Model Performance and Adoption
While the models demonstrate significant cost savings and maintained core capabilities, some uncertainties remain. It is not yet clear how these models will perform across diverse real-world applications, especially those requiring nuanced understanding or detailed outputs. The noted regressions in knowledge work and presentation quality suggest that certain workflows may need testing before full adoption. Additionally, long-term stability and performance under continuous use are still to be observed, as the models are relatively new.
Market response and competitive reactions are also evolving, making it uncertain how pricing strategies across the industry will shift in the coming months. The full impact on AI adoption rates and the competitive landscape remains to be seen.
Next Steps for Users and Industry Watchers
In the coming weeks, organizations should conduct thorough testing of GPT‑6 Sol and Luna within their specific workflows to assess performance and suitability. OpenAI is expected to release further tools and dashboards to optimize caching and effort management, aiding developers in cost control. Monitoring industry responses, including potential price adjustments from competitors, will be key to understanding market dynamics.
Additionally, ongoing evaluations of the models’ performance in real-world tasks will inform their broader deployment. OpenAI may also release updates or tuning options to address identified regressions, particularly in detailed output quality. Overall, the focus will be on balancing cost, quality, and reliability to maximize the models’ utility across sectors.
Key Questions
How much cheaper are GPT-6 Sol and Luna compared to previous models?
GPT‑6 Sol costs approximately $2.00 per 1 million tokens for input and $10.00 for output, while Luna costs $0.10 for input and $0.50 for output, roughly 50-60% less than GPT‑5.6 models.
Do the new models perform worse than their predecessors?
Independent analysis shows performance remains comparable, with some regressions in detailed knowledge tasks. Hallucination rates have improved, reducing errors significantly, but some quality aspects in detailed outputs have declined slightly.
What are the main benefits of these price reductions?
The primary benefit is increased accessibility and affordability, enabling more organizations to automate tasks and deploy AI at scale without high costs. This could accelerate AI adoption across industries.
Are there any trade-offs or downsides to the new models?
Yes, some regressions in detailed output quality and presentation have been observed, which might affect workflows requiring comprehensive, polished deliverables. Users should evaluate models carefully before full deployment.
What does this mean for the AI industry overall?
The price cuts signal a shift towards more competitive, accessible AI markets, potentially prompting other providers to lower their prices and innovate further in cost efficiency.
Source: ThorstenMeyerAI.com
