TechCrunch: OpenAI launches GPT-5.6 with three new model variants

On July 9, 2026, OpenAI released GPT-5.6, a new model family with three variants designed for different use cases and cost levels. The release follows a pattern the company has used to expand access — a premium model with maximum capability, a mid-range option, and a budget tier.

Sol is the premium variant, positioned as the highest-performance option for complex tasks. OpenAI describes it as the best coding model in the family and notes that it is 54% more token-efficient than predecessor models for coding tasks. Pricing is $5 per million input tokens and $30 per million output tokens.

Terra is the mid-range option, priced at $2.50 input and $15 output per million tokens.

Luna is the budget-tier variant at $1 input and $6 output per million tokens.

The announcement also included ChatGPT Work, a workplace-oriented product described as an agent that can act across files and applications to complete multi-step projects. This is separate from the model release itself but shipped alongside it.

For writers and content teams, the release is relevant on two levels. First, the pricing structure makes GPT-5.6 viable for higher-volume editorial workflows — teams that run large batches of drafts, summaries, or translations will see meaningfully lower costs per output at the Luna and Terra tiers. Second, Sol’s token-efficiency improvements affect response quality when working with long documents, which is relevant for research-heavy writing tasks.

OpenAI also described Sol as its “strongest cybersecurity model yet,” with applications in threat modeling and code review. That framing is aimed at enterprise buyers rather than editorial teams, but it signals that the GPT-5.6 family is positioned for professional and institutional use at a scale that the previous GPT-5 generation was not.