OpenAI has released three new reasoning models—Sol, Terra, and Luna—each optimized for different use cases: Sol for cybersecurity tasks, Terra for general-purpose reasoning comparable to GPT-5.5, and Luna for speed and cost efficiency.
The models are available in preview with tiered API pricing, allowing developers to choose the right trade-off between performance, capability, and expense for their needs.
What happened
OpenAI introduced three new AI models — Sol, Terra, and Luna — each designed for different tasks. Sol handles secure coding and cybersecurity work, Terra balances performance with GPT-5.5-level capability, and Luna prioritizes speed and low cost. All three are available in preview, with API pricing ranging from 1 token to 100 tokens per input and 6 tokens to 30 tokens per output, depending on the model.
Why it matters
The three models give developers and businesses a choice between speed, safety, and raw capability rather than a one-size-fits-all option. Organizations can now pick Sol for security-sensitive work, Terra for general-purpose reasoning, or Luna for cost-conscious applications — reducing unnecessary spending on more powerful models than needed.
What to watch
The models are currently in preview. Anthropic separately launched Claude Fable 5 in June and then halted development, restarting on July 1.
Ask the AI about this article →
OpenAI's release of three distinct reasoning models signals a shift from monolithic AI products toward specialized, use-case-optimized offerings. Rather than forcing all developers onto a single powerful (and expensive) system, the company is fragmenting its reasoning layer by speed, safety, and cost profile. This reflects the broader industry trend of moving beyond one-size-fits-all LLMs toward targeted deployments.
The pricing structure underscores the differentiation: Sol's 5-dollar input and 30-dollar output cost, Terra's middle-ground pricing, and Luna's 1-dollar and 6-dollar tiers create clear economic incentives for developers to match model to workload. Simultaneously, Anthropic's halt and restart of Claude Fable 5 development (paused after June, resumed July 1) suggests competitive pressure in the reasoning model space, with multiple vendors experimenting with tiered capability and cost strategies to capture different market segments.
For example, today's edition would include:
AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.
Free · takes 30 seconds · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. Q&As are published on this page for other readers too.
Google DeepMind chief Koray Kavukcuoglu said being at the frontier of AI is the only thing that matters to the…
John Deere is testing an AI assistant called “JD” that answers farmers' questions on topics like equipment set…
Google has launched Google Pics, a new suite of creative design tools for Workspace users, built around Gemini…
OpenAI said today that it is integrating ChatGPT Health with Epic's electronic health record (EHR) system, whi…
Google is launching Google Pics, an AI-powered image creation and editing tool that will be part of Google Wor…
Google DeepMind launched agentic video understanding across Gemini 3.7 Flash, 3.6 Flash, and 3.5 Flash-Lite
