OpenAI Launches GPT-6 Sol and GPT-6 Luna, Cuts API Prices to Half of GPT-5.6 Promotional Price
AI Podcast
OpenAI has launched the GPT-6 Sol and Luna models, aiming to provide affordable options for daily office, programming, and computer operation tasks. Sol targets complex workflows with high usage limits and superior automation performance, while Luna serves as a lightweight, budget-friendly choice for high-volume tasks. Both models significantly reduce API pricing and operational costs compared to previous generations, while substantially lowering error rates. Meanwhile, the release preceded Anthropic’s launch of Claude Opus 5.5 by an hour, highlighting intensifying market competition in cost-effective, high-performance artificial intelligence solutions.
TradingKey - OpenAI has launched GPT-6 Sol and GPT-6 Luna with a straightforward goal: to make the GPT-6 series affordable for more users. While GPT-6 Astra remains the most powerful model, Sol and Luna will handle daily office, programming, and computer operation tasks at lower prices. OpenAI stated that the new models focus on optimizing speed and cost, cutting API prices to half of GPT-5.6's promotional price.
Notably, about an hour before OpenAI released GPT-6 Sol, Anthropic launched Claude Opus 5.5.
What Is the Difference Between SOL and LUNA
GPT-6 Astra can be viewed as the "all-around flagship," GPT-6 Sol as the "high-performance work version," while GPT-6 Luna is better suited for high-volume, frequent everyday tasks.
Sol is primarily geared toward complex tasks. It can handle business workflows, long-duration coding, and computer operations, while offering users higher usage limits. Luna is positioned as a lighter option, making it suitable for budget-conscious individuals and teams with high call volumes.
Test results released by OpenAI show that Sol achieved a top score of 56.4% on AutomationBench, with costs 60% lower than Claude Opus 5 when completing similar tasks. In the same type of work, Luna improved by 5.4 percentage points compared to the previous generation, while the cost per task dropped by 58%.
Programming Costs Fall Significantly
For developers, price changes are particularly important. Today's coding assistants do not just write a few lines of code; they also inspect entire projects, modify multiple files, and even work continuously for hours. The more powerful the model and the more complex the task, the easier it is for costs to accumulate.
OpenAI stated that GPT-6 Sol scored 68.8% on the DeepSWE benchmark, just 1.1 percentage points behind Claude Fable 5's 69.9%, but with a per-task cost about 80% lower. GPT-6 Luna scored 66.6%, performing close to the medium settings of some competing models, yet with costs more than 93% lower.
More Accurate Answers That Also Save Money
One of the most concerning issues with AI is that answers sound plausible yet get the facts wrong. OpenAI stated that in internal real-conversation testing, GPT-6 Sol made about half as many errors as the previous generation, with reliability approaching that of GPT-6 Astra.
GPT-6 Luna has also made marked progress. Under higher workloads, it can achieve performance comparable to GPT-5.6 Sol, but at about one percent of the latter's cost.
This content was translated using AI and reviewed for clarity. It is for informational purposes only.
Recommended Articles
Comments (0)
Click the $ button, enter the symbol, and select to link a stock, ETF, or other ticker.