--- title: "OpenAI Launches GPT-6 Sol and Luna at Half the API Price" date: 2026-09-22 author: "Barry Elad" featured_image: "https://sqmagazine.co.uk/wp-content/uploads/2026/09/openai-gpt-6-sol-luna-half-price-api.jpg" categories: - name: "Artificial Intelligence" url: "/artificial-intelligence.md" tags: - name: "News" url: "/tag/news.md" --- # OpenAI Launches GPT-6 Sol and Luna at Half the API Price OpenAI released GPT-6 Sol and GPT-6 Luna on 22nd September, 2026, cutting API prices for both models by 50% against GPT-5.6 promotional pricing. The launch landed about 90 minutes after Anthropic shipped a new version of Opus 5.5. ## Key Takeaways - OpenAI added GPT-6 Sol and GPT-6 Luna as faster, cheaper members of the GPT-6 family that Astra opened earlier this month. - API prices for Sol and Luna run 50% below GPT-5.6 promotional pricing, a cut OpenAI credits to better caching and inference. - GPT-6 Sol makes about half as many mistakes as its predecessor on OpenAI’s internal factuality test, the company says. - Paid ChatGPT Work and Codex accounts get both models now. Free and Go users can use Luna in the desktop app. - Anthropic released a new Opus 5.5 shortly before OpenAI’s announcement, turning launch day into a head-to-head. ## What We Know? OpenAI announced the models in a [launch post](https://openai.com/index/introducing-gpt-6-sol-and-luna/) and on X. Both sit below **GPT-6 Astra**, which the company called its most capable model yet when it arrived earlier this month. Sol handles complex work such as coding. Luna targets what OpenAI describes as high-volume tasks with a clear goal, like summarizing documents, extracting information, or answering quick questions. The first [Sol and Luna models debuted in July](https://sqmagazine.co.uk/openai-launches-gpt-5-6-sol-new-ai-features/) as part of the GPT-5.6 series. Today’s versions carry over much of Astra’s strength into smaller, faster tiers, according to OpenAI. This is the second GPT-6 launch this month. Both models are live now in ChatGPT Work and Codex for most paid accounts, plus the API. OpenAI expects the ChatGPT app and website to receive them gradually throughout the day. > Please welcome GPT-6 Sol and GPT-6 Luna to the GPT-6 universe. > > GPT-6 Sol and Luna build on the advances behind GPT-6 Astra, bringing much of its strengths into faster and more affordable models to support work at scale. > > We’ve also made caching and inference more efficient, and… [pic.twitter.com/5LiVE4rbFt](https://t.co/5LiVE4rbFt) > > — OpenAI (@OpenAI) [September 22, 2026](https://x.com/OpenAI/status/2102460975790137662?ref_src=twsrc%5Etfw) ## Half the Price, Fewer Mistakes “**We’ve also made caching and inference more efficient, and we’re passing the savings directly to you: 50% lower API prices for Sol and Luna compared with GPT-5.6 promotional pricing**,” OpenAI said on X. The baseline matters. The cut is measured against promotional GPT-5.6 rates, so teams should compare it with what they actually paid. On accuracy, OpenAI says its internal factuality evaluation uses de-identified real-world conversations in which users flagged model mistakes. By that measure, GPT-6 Sol makes about **half** as many mistakes as its predecessor, “**approaching Astra-level reliability at much lower cost.**” The company also reports a lower error rate on coding. Those numbers come from OpenAI alone. The test counts mistakes users noticed and flagged, so it cannot capture errors nobody caught. No independent reproduction accompanied the launch. ## What We Don’t Know? OpenAI claims Sol and Luna handle tasks substantially better than Anthropic’s top models, including Fable and Opus. Several questions remain open: - **Did OpenAI’s comparisons include the Opus 5.5 update that Anthropic shipped just before?** - **Will the factuality gains hold on independent benchmarks?** - **How does the price cut look against standard GPT-5.6 list prices?** - **When will every ChatGPT account see the new models?** [Anthropic’s commercial scale](https://sqmagazine.co.uk/claude-ai-statistics/) is why those head-to-head claims carry weight. ## What’s Next? Price is the sharper weapon in this release. **Sol and Luna** serve coding and clerical workloads, where token bills grow fastest at scale. Halving those costs while claiming fewer errors pressures Anthropic on the tiers built for everyday, high-volume work. The near simultaneous launches show how tightly the rivals now shadow each other. **What comes next is testing**. The ChatGPT rollout should finish over the day, and outside benchmarks pitting Sol against [Opus 5.5](https://sqmagazine.co.uk/anthropic-claude-opus-5-5-launch/) are the obvious next data point. Developers already on GPT-5.6 Sol or Luna should rerun their own evaluations and check cached-prompt costs before moving production traffic. Vendor scores rarely map cleanly onto a specific workload. Definition of AI Inference. Link to full glossary entry follows the description.**AI Inference**AI inference is the execution phase where a trained AI model applies what it learned to new, unseen data and produces an output such as a prediction. [Read more](https://sqmagazine.co.uk/glossary/ai-inference/)