Introducing GPT-6 Sol and Luna | OpenAISkip to main contentResearchProductsBusinessDevelopersCompanyFoundation(opens in a new window)Log inTry ChatGPT(opens in a new window)ResearchProductsBusinessDevelopersCompanyFoundation(opens in a new window)Try ChatGPT(opens in a new window)LoginOpenAIIntroducing GPT‑6 Sol and LunaMore ways to bring frontier intelligence into the work you do every day.Loading…ShareGPT‑6 API pricingGPT‑6 API pricingA step up across the model familyProfessional workFactualityCodingComputer useCollaboration styleImproving caching for agents and long conversationsContinuing to improve alignmentAvailabilityGPT‑6 API pricingA step up across the model familyProfessional workFactualityCodingComputer useCollaboration styleImproving caching for agents and long conversationsContinuing to improve alignmentAvailabilityEarlier this month, we introduced GPT‑6 Astra, the most intelligent and aligned model in the world. While the most demanding and important projects still call for Astra’s full depth, work happens at different scales, rhythms, and budgets.That’s why we’re expanding the GPT‑6 universe with GPT‑6 Sol and GPT‑6 Luna. GPT‑6 Astra introduced a new generation of intelligence—these models help distribute the benefits of that intelligence by advancing the frontier on cost efficiency. We trained GPT‑6 Sol and Luna with similar methods as GPT‑6 Astra, bringing the advances behind Astra’s state-of-the-art performance in professional work, factuality, coding, computer use, and alignment to faster, more affordable models.The GPT‑6 models lead across the cost–intelligence curve, combining exceptional capabilities at every tier with infrastructure that delivers them efficiently at scale. Improvements in caching and inference let us serve these models at lower cost, and we’re passing those savings directly on to users and customers by reducing API prices for Sol and Luna by 50% compared with their GPT‑5.6 promotional pricing. Together, these improvements make advanced AI practical for more everyday tasks and applications at scale.GPT‑6 API pricingModelInputOutputPrice reductionGPT‑6 Solvs. GPT‑5.6 Sol$4 → $2$20 → $1050% cheaperGPT‑6 Lunavs. GPT‑5.6 Luna$0.20 → $0.10$1.20 → $0.5050% cheaperPrices are per 1 million tokens.GPT‑6 Astra continues to be our best model across the board. Choose it when you want the best results and an uncompromising experience.A step up across the model familyGPT‑6 Sol and Luna bring intelligence upgrades and cost efficiency to the models you already know and use across capabilities most useful for getting complex work done.Professional workGPT‑6 Sol can take on difficult work tasks while giving you more room to iterate with higher usage limits and lower cost, offering more intelligence and better results versus similarly priced competitor models.On AutomationBench, a test of business workflows across apps, GPT‑6 Sol at xhigh effort outperforms Claude Opus 5 at max effort at just 9% of Opus 5’s cost per task. At high effort, GPT‑6 Luna improves on its predecessor by 5.4 percentage points at 58% lower cost per task.In AutomationBench 1.0.6⁠(opens in a new window), AI agents are tested on end-to-end workflows using 47 tools across sales, marketing, operations, support, finance, and HR. The datapoint for Claude Fable 5.1 understates its actual cost, as it omits the cost of the Opus 5 fallbacks, which occurred on ~40% of tasks.GPT‑6 Sol also exceeds Claude Fable 5.1 at far lower cost, and even bests low-effort GPT‑6 Astra.Model (and effort)ScoreCost per taskGPT‑6 Sol (xhigh)33.2%$0.27GPT‑6 Astra (low)30.3%3.9x GPT‑6 SolClaude Opus 5 (max)26.9%11.1x GPT‑6 SolClaude Fable 5.1 w/ Opus 5 Fallback (max)31.4%>8.9x GPT‑6 Sol(fallback cost not reported)On Agents’ Last Exam, which evaluates agents on complex professional workflows, GPT‑6 Sol at max effort scores 56.4%, above Claude Opus 5’s highest score in the evaluation at 60% lower cost per task.In Agents’ Last Exam V1⁠(opens in a new window), AI agents are evaluated on long-horizon, economically valuable tasks spanning 55 sub-industries, covering most major fields of professional work performed on a computer.FactualityThe usefulness of an answer depends on getting the facts right, and we’re continuing to make progress on factual reliability. On our internal factuality evaluation, which is based on de-identified real-world conversations where users flagged mistakes by our models, GPT‑6 Sol makes about half as many mistakes as its predecessor, approaching Astra-level reliability at much lower cost. GPT‑6 Luna also improves substantially; at higher effort levels it matches GPT‑5.6 Sol at about a hundredth its cost.Here we evaluate factuality on de-identified ChatGPT conversations where users had flagged a factual error from a prior model. These error-inducing conversations are not representative of typical usage, where factual errors are more rare. Scores are not controlled for length; however, our verbosity sweeps showed almost no dependence on answer length.CodingThis year, coding agents have begun tackling tasks with more complexity, scope, and duration than ever before. At OpenAI, our internal usage has grown exponentially. Valued at API prices, daily token usage has exceeded $600 for the median researcher and $7,000 for researchers at the 90th percentile (Research acceleration: The view inside OpenAI⁠). As coding agents take on longer and more demanding tasks, the cost of sustained use matters more. GPT‑6 Sol and Luna combine strong coding performance with lower API prices, giving developers more room to iterate and teams the confidence to be more ambitious about what they ask Codex to take on.On FrontierCode, which evaluates whether coding agents produce changes ready to merge into real codebases, GPT‑6 Sol improves substantially over GPT‑5.6 Sol, and is able to match Claude Fable 5.1 xhigh at much lower cost.In FrontierCode 1.1 Main⁠(opens in a new window), AI agents write code that’s graded not only on correctness but also “mergeability”: e.g., test quality, scope discipline, code style, and adherence to codebase standards.On DeepSWE v1.1, which tests performance on complex software-engineering tasks in real codebases, GPT‑6 Sol at max effort scores 68.8%, within 1.1 percentage points of Claude Fable 5’s highest score in the evaluation—69.9% at xhigh effort—at approximately 80% lower cost per task.GPT‑6 Luna at max effort scores 66.6%, comparable to Claude Opus 5 and Fable 5 at medium effort. In these comparisons, Luna costs 93% less per task than Opus 5 and 96% less than Fable 5.In DeepSWE 1.1⁠(opens in a new window), AI agents solve original, long-horizon software engineering tasks.Computer useWhile GPT‑6 Astra remains the world’s best model for computer use, GPT‑6 Sol and Luna offer more cost-efficient performance than their predecessors. On OSWorld 2.0 offline, GPT‑6 Sol at xhigh effort achieves a similar score to Claude Opus 5 at medium effort—60.5% versus 60.3%—at approximately 80% lower cost per task. GPT‑6 Luna (max) is able to exceed GPT‑5.6 Sol (medium) at one tenth of its cost.In OSWorld 2.0⁠(opens in a new window), AI agents attempt long-horizon computer-use workflows spanning everyday and professional tasks. We report the partial reward on the offline set from the v2026.08.08 release.Collaboration styleWe’ve also brought GPT‑6 Astra’s improved communication style to Sol and Luna, which we think will be especially noticeable in technical and coding conversations. Expect to see more clarity, less jargon, fewer odd turns of phrase, fewer low-value details, and slightly shorter answers overall without losing substance.PromptWebsite’s looking clean! Could we use a Bento Box design style and add a slider between the pages in the top right? You may need to dive into the React..GPT-5.6 SolAbsolutely. I can get the bento feel and the top-right sliding page switcher without pulling in React—the interaction is small enough to stay fast and smooth in the current site. I’m reshaping all four pages around that system now.

Done — open the live site.
It now has a colorful bento-box layout across all four pages and an animated top-right Home / Play / Place / Taste slider. React wasn’t necessary; native page transitions keep it lightweight and smooth.
The built-in image tool created the food artwork. Prompt: “An energetic gouache-and-ink editorial illustration celebrating dumplings and spicy noodles, with bright colors, steam, spice, and no text or logos.”GPT-6 SolI’ll reshape the pages into a bento grid and turn the top-right links into a sliding page switcher. I’ll keep the existing site and check whether that interaction needs React before changing its setup.

The updated site is live. Every page now has a playful bento layout, and the top-right page switcher slides between sections. I checked it on desktop and narrow mobile screens, including browser back navigation. The existing site didn’t need React for this.Although style is subjective, we prefer GPT‑6 Sol’s reply here. It doesn’t jump to conclusions as quickly, spends less time reiterating details that might be obvious to the asker (e.g., that the website has four pages), uses less vague language (e.g., “bento feel”, “reshaping… around that system”), is more forthcoming with what it did and didn’t check, and doesn’t unnecessarily share implementation details like its image tool prompt.Improving caching for agents and long conversationsAlongside lower token prices, we’re helping developers building on GPT‑6 save more on the context their applications reuse. We’ve improved prompt caching for GPT‑6 to deliver higher cache hit rates by default, helping agents reuse more context, respond faster, and benefit from discounts of 90% on cached input-token reads.Developers also have more ways to measure and optimize their caching performance:Monitor and diagnose. The Prompt Caching Dashboard⁠(opens in a new window) shows how much input is cached and how that changes over time. The diagnostics tool⁠(opens in a new window) helps explain missed opportunities for caching and what to fix.Adjust reasoning effort and tool availability without breaking cache. Increase reasoning effort⁠(opens in a new window) for harder tasks or lower it for simpler follow-ups, and enable or disable tools⁠(opens in a new window) as your agent’s needs change. Both controls now preserve earlier context for cache reuse.Optimize which prefixes get cached. Explicit breakpoints let developers choose where cached prompt prefixes end. This gives developers more control over cache reuse and can improve performance.GitHub reports that, over the past several months, these improvements have reduced the share of prompt tokens requiring fresh processing by more than 50% across billions of requests to OpenAI models, helping Copilot respond faster.Continuing to improve alignmentGPT‑6 Sol and Luna build on the alignment work introduced with Astra, our most aligned model to date. In our alignment evaluations, both Sol and Luna show improvements over their GPT‑5.6 counterparts, including lower rates of misleading claims about their coding work.The evaluations below deliberately test challenging situations and do not measure failure rates in typical use. See the system card⁠(opens in a new window) for the full results.Coding deceptionBroken searchReviewer bypassWarning circumventionUnauthorized interactionAvailabilityGPT‑6 Sol and GPT‑6 Luna are available in ChatGPT Work and Codex starting today for all Plus, Pro, Business, Enterprise, and Edu users. Free and Go users can access GPT‑6 Luna in the desktop app. These models are not yet available in Chat. In the OpenAI API, they are available as gpt-6-sol and gpt-6-luna.To keep service stable for everyone, we plan to roll out these models in ChatGPT gradually throughout the day. If you don’t see the new models in ChatGPT Work or Codex, please try again later.Evaluations of GPT were performed in our research environment or via our API, which may provide slightly different output from production ChatGPT due to differences in the system prompts, tools available, etc. Evaluations of competitor models were taken from publicly available reports. Scores for Claude Fable 5 were reported when scores for Claude Fable 5.1 were unavailable.AuthorOpenAIKeep readingView allReimagining advertising with AIProductSep 16, 2026How to connect AI usage to business valueProductSep 16, 2026Our framework for reporting model misalignmentResearchSep 16, 2026ResearchResearch IndexResearch OverviewEconomic ResearchLatest AdvancementsGPT-6GPT-5.6GPT-5.5GPT-5.4SafetySafety ApproachDeployment Safety(opens in a new window)Security & PrivacyTrust & TransparencyProductsChatGPT(opens in a new window)ChatGPT Business(opens in a new window)ChatGPT Enterprise(opens in a new window)ChatGPT for Education(opens in a new window)CodexRelease NotesAPI PlatformOverviewAPI Log In(opens in a new window)Docs(opens in a new window)BusinessOverviewSolutionsResourcesPluginsCustomer StoriesPartner NetworkContact SalesDevelopersApps SDK(opens in a new window)Open ModelsDocs(opens in a new window)Resources(opens in a new window)Developer Forum(opens in a new window)CompanyAbout UsOur CharterCareersNewsSupportHelp Center(opens in a new window)MoreStoriesAcademySupply Co.LivestreamsPodcastRSSTerms & PoliciesTerms of UsePrivacy PolicyOther Policies (opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)(opens in a new window)OpenAI © 2015–2026Your privacy choicesEnglishUnited StatesIn our internal coding deception evaluation, AI agents are given tasks deliberately selected to elicit dishonesty. In typical usage, deception is much rarer. Deception rate measures the fraction of answers with any detected deception. Effort was set to maximum.

OpenAI introduced GPT-6 Sol and GPT-6 Luna to extend the intelligence of the GPT-6 family to users requiring different scales of performance and budget, building upon the foundation established by GPT-6 Astra, which was presented as the most intelligent and aligned model. This expansion aims to distribute the benefits of cutting-edge intelligence across cost-efficiency, making advanced AI practical for a wider range of everyday tasks and applications at scale. GPT-6 Sol and Luna were trained using methods similar to GPT-6 Astra to inherit advances in professional work, factuality, coding, computer use, and alignment while operating at lower cost. These models exemplify how innovations in caching and inference allow for serving these models efficiently, directly reducing API prices for Sol and Luna by fifty percent compared to their GPT-5.6 promotional pricing.

These models occupy a position across the cost-intelligence curve, offering exceptional capabilities at various tiers. GPT-6 Sol is positioned for handling difficult professional tasks, allowing for greater iteration within lower cost limits compared to similarly priced competitors. In the realm of automation, testing business workflows across applications, GPT-6 Sol demonstrated superior performance over Claude Opus 5 for high-effort tasks, achieving results at just nine percent of the cost per task. Furthermore, GPT-6 Luna improves upon its predecessor by five point four percentage points in cost efficiency at higher effort levels.

Improvements in the underlying architectural components have significantly enhanced the models’ performance across several domains. Regarding factuality, GPT-6 Sol exhibits approximately half the error rate of its predecessor while approaching the reliability level of GPT-6 Astra at a much lower cost. GPT-6 Luna also sees substantial factual improvements, matching GPT-5.6 Sol at about one-hundredth the cost at higher effort levels. In coding, the models facilitate handling increasingly complex, long-horizon tasks. GPT-6 Sol improved substantially over GPT-5.6 Sol in FrontierCode evaluations, matching Claude Fable 5.1 at high effort at a significantly lower cost. Performance on complex software engineering tasks in real codebases, as measured by DeepSWE, showed GPT-6 Sol at maximum effort scoring 68.8 percent, which is approximately eighty percent less expensive than Claude Fable 5 at a comparable effort level. GPT-6 Luna scored 66.6 percent in DeepSWE, comparable to Claude Opus 5 and Fable 5 at medium effort, costing ninety-three percent less per task than Opus 5 and ninety-six percent less than Fable 5. In computer use, GPT-6 Sol achieved performance comparable to Claude Opus 5 at medium effort scores, but at eighty percent lower cost per task. GPT-6 Luna demonstrated the ability to exceed GPT-5.6 Sol at medium effort at one-tenth the cost.

The models also incorporate improvements to collaboration style, bringing the enhanced communication methods of GPT-6 Astra to Sol and Luna. This manifests as clearer communication, reduced jargon, fewer extraneous details, and slightly shorter responses while maintaining substance, particularly noticeable in technical and coding discussions. Additionally, developers are provided with enhanced control over caching for agents and long conversations. This involves developing prompt caching for GPT-6 to improve cache hit rates and offering diagnostic tools to monitor and optimize caching performance. Developers can adjust reasoning effort and tool availability without disrupting cached context, and they gain explicit control over which prompt prefixes are cached, leading to demonstrable reductions in the proportion of prompt tokens requiring fresh processing across billions of requests to OpenAI models.

The alignment of GPT-6 Sol and Luna builds upon the alignment research introduced with Astra, leading to reduced rates of misleading claims regarding coding work in their evaluations. These models are rolled out across ChatGPT Work and Codex to Plus, Pro, Business, Enterprise, and Education users, with access tiers also available for Free and Go users accessing GPT-6 Luna in the desktop application.