🔍 Read the full analysis: OpenAI’s Reduced Prices For GPT‑6 Sol And Luna Don’t Impact Benchmarks on ThorstenMeyerAI.com
Get the latest gadgets 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 launched GPT‑6 Sol and Luna at half the price of previous models, yet independent evaluations indicate their benchmark scores and capabilities remain stable. Cost savings do not translate into performance improvements, but they could expand AI deployment options.
OpenAI has released GPT‑6 Sol and GPT‑6 Luna at **50% lower prices** than their GPT‑5.6 predecessors, aiming to make advanced AI more accessible without sacrificing performance. The models arrived on September 22, 2026, and the company emphasizes that these price reductions are achieved through improvements in caching and inference efficiency, not by altering the models’ capabilities. Independent benchmark evaluations confirm that **model scores and performance metrics remain roughly consistent** with previous versions, despite the significant cost cuts.
On September 22, 2026, OpenAI announced the launch of **GPT‑6 Sol and GPT‑6 Luna**, with prices cut by **50%** compared to GPT‑5.6 models. The new pricing reflects a strategic shift toward **cost-efficient AI deployment**, enabling broader use cases in commercial and operational workflows. The models’ input and output token costs are now $2.00 and $10.00 per million tokens for Sol, and $0.10 and $0.50 for Luna, representing a substantial reduction from previous rates.
OpenAI attributes these savings to **improvements in caching and inference techniques**, which allow the models to be served at lower costs. For example, cached input reads now receive **90% discounts**, and caching-related premiums are maintained at 25%. The company states that these efficiencies do not compromise the models’ core capabilities, which are focused on delivering high-quality outputs at a lower price point.
Independent analysis published on the same day by Artificial Analysis confirms that **cost per task has roughly halved** while **model scores and performance metrics remain stable**. The analysis reports that GPT‑6 Sol at maximum effort costs about **$1.06 per task**, compared to **$1.99 for GPT‑5.6 Sol**, and Luna costs approximately **$0.07 per task**, down from previous costs. Despite spending slightly more output tokens per task, the savings are primarily driven by **price reductions rather than efficiency gains**. The models scored well on benchmarks, with Sol achieving a score of **48 on the Artificial Analysis Intelligence Index**, well above the median of 25, and Luna scoring 37 against a median of 12.
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 Lower Prices on AI Deployment and Benchmarks
The release of **more affordable GPT‑6 models** could significantly **expand AI adoption** across industries by lowering operational costs. While **performance remains stable**, the reduced prices make it feasible for organizations to automate more tasks, especially in customer service, research, and content generation. However, the **performance benchmarks show no improvement** over previous models, indicating that the core capabilities are maintained rather than enhanced. This shift could **reshape market dynamics**, emphasizing cost-efficiency over model sophistication, and potentially **accelerate AI integration** into everyday workflows.
As an affiliate, we earn on qualifying purchases.
Background on GPT Model Pricing and Performance Trends
OpenAI’s previous models, including GPT‑5.6, set the benchmark for AI capabilities and pricing. Historically, improvements in model architecture and training have driven performance gains, often accompanied by increased costs. The recent introduction of GPT‑6 Sol and Luna marks a **departure from this trend**, focusing on **cost reduction** without sacrificing core performance metrics. The models are part of OpenAI’s broader strategy to **democratize access** to advanced AI by making it more affordable for a wider range of users and applications.
Independent evaluations, such as those by Artificial Analysis, have historically tracked the relationship between model capabilities and cost, providing a benchmark for assessing whether price cuts come at the expense of quality. Prior to this release, models like Claude Opus 5.5 also reduced prices but achieved only modest gains in performance. The current models’ performance stability, despite significant price cuts, suggests a **mature stage in model optimization** where cost-efficiency is prioritized.
As an affiliate, we earn on qualifying purchases.
Unconfirmed Aspects of Model Performance and Market Impact
While independent benchmarks confirm performance stability, it is **still unclear how these models will perform across diverse real-world applications** beyond standardized tests. Additionally, the long-term impact on **market competition and AI pricing trends** remains uncertain**, as other providers may respond with their own cost strategies. The full extent of how these price reductions will influence **adoption rates and user behavior** is also yet to be seen.
As an affiliate, we earn on qualifying purchases.
Future Developments and Market Response to Price Cuts
OpenAI is expected to **continue refining** its models and caching techniques to further reduce operational costs. The company may also **expand the deployment** of GPT‑6 Sol and Luna in commercial products, with user feedback guiding future tuning. Industry analysts predict that **competitors will monitor** these developments closely, possibly leading to **further price adjustments** across the AI market. Monitoring how these models perform in real-world settings over the coming months will clarify their practical impact.
As an affiliate, we earn on qualifying purchases.
Key Questions
Do the lower prices mean the models are less capable?
No, independent benchmarks confirm that **performance scores and capabilities remain stable** despite the price reductions. The models are optimized for cost, not capability.
Will the price cuts lead to broader AI adoption?
Yes, the significant reduction in costs could enable **more organizations to deploy AI solutions**, especially in areas where budget constraints previously limited adoption.
Are there trade-offs in quality or accuracy?
Some regressions in specific knowledge tasks have been observed, likely due to tuning for cost efficiency. However, **core capabilities and benchmark scores remain consistent**, with improvements mainly in hallucination reduction and refusals.
How might competitors respond to this pricing strategy?
Competitors like Anthropic and others may **adjust their prices or model offerings** in response, potentially leading to a more competitive market focused on both cost and performance.
What should users test before switching to these models?
Users with workflows that involve **producing comprehensive, well-presented deliverables** should **test the models’ output quality** to ensure it meets their standards, as some regressions in presentation quality have been noted.
Source: ThorstenMeyerAI.com
Fall Picks
fall essentials
As an affiliate, we earn on qualifying purchases.
