OpenAI has begun rolling out GPT-6 Sol and GPT-6 Luna to Codex and ChatGPT Work at lower token prices than their GPT-5.6 predecessors1. The rollout, logged on September 22, splits the new generation into two models tuned for distinct workloads: GPT-6 Sol targets complex coding and agentic workflows, while GPT-6 Luna handles focused, high-volume tasks.
Both models are available to Plus, Pro, Business, Enterprise, and Edu users. Free and Go users can access GPT-6 Luna through the desktop app. Enterprise administrators must explicitly enable the new models; availability also depends on rollout status and workspace settings.
Within ChatGPT, GPT-6 Sol and GPT-6 Luna appear in Work and Codex but not in Chat. Users can select a model through the picker in the interface or, in the CLI, via the `/model` command or by launching a task with `codex --model gpt-6-sol` or `codex --model gpt-6-luna`.
The changelog states that both models carry lower token prices than GPT-5.6. ANALYSIS Pairing a generational model upgrade with a price cut below the prior release inverts the typical pattern in which newer, more capable models command a premium at launch. The two-model split also formalizes a tiering strategy: GPT-6 Sol as the heavyweight for multi-step agent tasks, GPT-6 Luna as the leaner option for throughput-sensitive pipelines.
The rollout arrives days after researchers disclosed Plugin4Shell, a zero-click remote code execution vulnerability affecting Codex and three other major AI coding agents[3]. ◆ Enterprise administrators weighing whether to enable GPT-6 Sol and GPT-6 Luna in Codex will need to account for both the capability upgrade and the security posture of the agent environment.
OpenAI's changelog directs users to its Models page for selection and migration guidance, its Pricing page for credit rates and usage details, and its Workspace model availability documentation for administrator controls.