The Logic Behind OpenAI’s Price Cuts And Unchanged Benchmarks For GPT‑6 Sol And Luna
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: The Logic Behind OpenAI’s Price Cuts And Unchanged Benchmarks For GPT‑6 Sol And Luna on ThorstenMeyerAI.com

Prime Big Deal Days · Oct 6–7Offer from Amazon

Get privacy and security 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
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

TL;DR

OpenAI has announced a 50% price reduction for its GPT-6 Sol and Luna models, achieved through improved caching and inference. Despite lower costs, benchmarks remain stable, though some regressions in knowledge tasks are reported. This shift makes AI more accessible for broader applications.

OpenAI has announced a 50% reduction in the prices for its GPT-6 Sol and Luna models, effective immediately. The models’ performance benchmarks remain largely unchanged, but the cost savings are driven by technological efficiencies in caching and inference. This move aims to make advanced AI more accessible for commercial and operational use, shifting the economics of deploying large language models.

On September 22, 2026, OpenAI released GPT-6 Sol and Luna, two models positioned as more cost-effective alternatives within their GPT-6 family. These models are priced at half the cost of their GPT-5.6 predecessors, with GPT-6 Sol at $2.00 per 1 million tokens for input and $10.00 for output, and GPT-6 Luna at $0.10 and $0.50 respectively. The company attributes the price cuts to improvements in caching and inference technology, which reduce operational costs while maintaining model performance.

Independent evaluation by Artificial Analysis confirms that the cost per task has roughly halved, with minimal impact on the models’ intelligence scores. GPT-6 Sol scores 48 on the Artificial Analysis Intelligence Index, significantly above the median of 25 for comparable models, while Luna scores 37 against a median of 12. These models also show notable reductions in hallucination rates, with Sol’s hallucination rate dropping from 92% to 60%, and Luna’s from 93% to 77%. Despite these improvements, some regressions in knowledge tasks have been observed, suggesting the tuning for better conversational behavior may slightly impair factual accuracy in specific domains.

At a glance
updateWhen: announced September 22, 2026
The developmentOpenAI lowered the prices of GPT-6 Sol and Luna models by half while maintaining their performance benchmarks, driven by efficiency improvements, affecting AI deployment costs and strategies.

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.

GPT‑6 Sol
$4 / $20 → $2 / $10
GPT‑6 Luna
$0.20 / $1.20 → $0.10 / $0.50

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.

GPT‑5.6 Sol
$1.99
GPT‑6 Sol
$1.06
GPT‑5.6 Luna
$0.18
GPT‑6 Luna
$0.07

The effort dial moves cost more than the model choice

Model and effortIntelligence IndexCost 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

Already on GPT‑5.6 Sol or Luna? The move is mostly a price cut. Re‑test first if your output is a document someone reads, not data a system consumes.
Shelved an automation on cost? Token prices halved and the effort dial adds another order of magnitude. Re‑run the business case.
Choosing between labs? The question is no longer which model is smartest, but which clears your quality bar at the lowest cost per task.
ThorstenMeyerAI.comSources: OpenAI (pricing, vendor benchmarks) and Artificial Analysis (independent evaluation and model pages). Figures as of 23 September 2026.

Implications for AI Deployment and Cost Efficiency

This price reduction signifies a major shift in the economics of deploying large language models. By halving costs without sacrificing performance, OpenAI enables broader adoption across industries such as customer service, research, and automation. It lowers the barrier for companies to incorporate advanced AI, potentially accelerating innovation and operational efficiency. However, the noted regressions in some knowledge tasks highlight the importance of testing these models within specific workflows to ensure suitability, especially for tasks requiring high factual accuracy.

Amazon

AI development hardware

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Background on OpenAI’s Model Pricing and Performance

OpenAI’s GPT models have historically been priced based on their size and capabilities, with GPT-5.6 models costing significantly more. Recent developments, including the release of GPT-6 Astra and subsequent models, focus on balancing performance with cost. OpenAI has emphasized that technological improvements, especially in caching and inference, allow for these price cuts while maintaining benchmarks. The move follows industry trends where companies seek to democratize AI access by reducing operational costs, making advanced models viable for a wider range of applications.

Prior to this announcement, OpenAI’s models were considered premium tools primarily used in high-value sectors. The new models, with their lower price points, aim to expand usage to smaller companies and broader use cases, shifting the competitive landscape and setting new standards for cost-performance ratios in large language models.

Amazon

large language model API access

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Unconfirmed Aspects of Model Performance and Future Changes

It is not yet clear whether OpenAI plans further adjustments to these models or additional price cuts in the near future. The long-term impact of the observed regressions on knowledge tasks remains uncertain, and whether OpenAI will release subsequent updates to address these issues is still to be seen. Additionally, the full scope of how caching improvements translate into operational savings across diverse deployment scenarios is still being evaluated.

Amazon

AI inference optimization tools

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Next Steps for OpenAI and Model Adoption

OpenAI is likely to continue refining its models, balancing performance with cost efficiency. Customers should monitor upcoming updates and conduct testing within their workflows to ensure suitability. Industry analysts will also observe how competitors respond and whether similar cost reductions occur elsewhere, potentially reshaping the AI market landscape. Further transparency on long-term performance and tuning strategies may follow as the models are adopted more widely.

Amazon

AI model caching solutions

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

Why did OpenAI reduce the prices of GPT-6 Sol and Luna?

OpenAI achieved cost reductions through technological improvements in caching and inference, allowing them to lower operational expenses and pass savings to customers.

Do the lower-priced models perform worse than previous versions?

Performance benchmarks remain largely stable, though some knowledge tasks show regressions. Overall, the models maintain high scores in intelligence and coding tasks, with improved hallucination rates.

How might these price cuts impact AI adoption?

The reduced costs make advanced AI more accessible for a broader range of companies and applications, potentially accelerating deployment and innovation across industries.

Are there any trade-offs with the new models?

Yes, some regressions in knowledge accuracy and presentation quality have been observed, indicating a trade-off between conversational smoothness and factual precision.

What should users do before switching to these models?

Users should test the models within their specific workflows to assess suitability, especially for tasks requiring high factual accuracy or detailed output.

Source: ThorstenMeyerAI.com

FALL

Fall Picks

As an affiliate, we earn on qualifying purchases.

You May Also Like
Mobilisiert, nicht ausgegeben: Was von Europas €200-Milliarden-KI-Offensive übrig bleibt

Mobilisiert, nicht ausgegeben: Was von Europas €200-Milliarden-KI-Offensive übrig bleibt

The EU’s InvestAI plan is billed as €200B for AI, but most of the figure is mobilized capital, not direct spending, as a July tender nears.
The Model Is Only 10%: The Real Lesson of the New SDLC

The Model Is Only 10%: The Real Lesson of the New SDLC

A new Google whitepaper reveals that the core of AI-driven software development isn’t the model itself, but how developers structure, verify, and control AI outputs.
How SenseTime Continues To Lead In The AI Frontier In 2026

How SenseTime Continues To Lead In The AI Frontier In 2026

A 2026 analysis claims SenseTime is pursuing dominance as a frontier AI lab, but lacks supporting evidence. The story explores what is known and what remains uncertain.
IdeaClyst: The Engine That Decides What’s Worth Building

IdeaClyst: The Engine That Decides What’s Worth Building

IdeaClyst launches as an idea engine transforming rough concepts into validated, targeted product proposals by analyzing roadmaps and web opportunities.