OpenAI Launches GPT-5.6 and ChatGPT Work: What Changes for Your B2B Team

On July 9, OpenAI introduced three new models (Sol, Terra, and Luna) and an agent called ChatGPT Work that no longer just answers — it does the work. What this means for a B2B sales team and how to think about adoption without breaking your process.

On July 9, 2026, OpenAI made its most important launch of the year: a new GPT-5.6 model family with three tiers (Sol, Terra, and Luna) and an agent called ChatGPT Work that represents a fundamental shift — from a chat that answers questions to an agent that executes complete work.

The technical details are already in every release note. In this article, we care about the other question: if you have a B2B sales team, what do you do with this? We review each piece, its pricing, and how to think about adoption without wasting money or breaking processes that already work.

GPT-5.6: three models, three budgets

OpenAI adopted the tiered approach Anthropic had been using with Opus/Sonnet/Haiku: instead of a single model, three options that balance capability, speed, and cost. The logic is that you only use the most expensive one for what justifies it.

• $5 / $30, Sol: The most capable. USD per million tokens (input / output). For complex reasoning, multi-step planning, demanding analysis.
• $2.50 / $15, Terra: The balance. Performs competitively with GPT-5.5 at half the cost. Designed for high-volume everyday work.
• $1 / $6, Luna: The fastest and cheapest. For simple tasks, short answers, very high volume.

The takeaway for a B2B team is straightforward: if you've been using GPT-5.5 in production for reporting automation, lead classification, or email responses, switching to Terra cuts costs in half without lowering quality. It's the "cheap upgrade" of the month.

Practical allocation rule: Luna for classifying, summarizing, and high-volume short answers. Terra for 90% of everything else (writing messages, lead analysis, report generation). Sol only when you truly need deep thinking (automation architecture, complex commercial plan, difficult debugging).

ChatGPT Work: the agent that does, not answers

ChatGPT Work is a different piece entirely. It's an agent that takes scattered context (documents, Slack, Notion, Microsoft 365, Google Drive) and turns it into finished deliverables: a document, a spreadsheet, a presentation, an analysis. It doesn't give you steps to build it — it builds it.

For a B2B team, the most obvious use cases are sales month-end close (build the report from CRM + email + notes), meeting preparation (gather everything we know about the account and return a one-pager), and commercial proposals (structured draft from the discovery conversation). Administrative work that currently eats hours from the SDR or the sales manager.

How to think about adoption for your team

• Don't change everything at once. Pick one specific process (the weekly report, meeting follow-ups) and try ChatGPT Work there for a month before scaling.
• Migrate your existing GPT-5.5 automations to Terra as soon as you can — it's money left on the table every day.
• Reserve Sol for what truly needs it. If you're using the most expensive model for tasks Luna can handle, you're overpaying.
• Don't automate the close. The agent drafts; a person reviews before it goes to the client.

Beware of the "agent everything" trap: The temptation with ChatGPT Work is to have it do everything. In a B2B context, that's risky: a poorly sent message to a decision-maker can burn an entire account. Always keep a human reviewing what goes out. The agent saves time; it doesn't replace commercial judgment.

What to watch in the next 30 days

GPT-5.6 and ChatGPT Work launched with limited preview and some US Gov restrictions, but they'll expand fast. What's worth watching: whether your stack (Notion, Slack, Google Workspace, your CRM) gets native integration, and how Terra performs in your real use cases — not in benchmarks.

Conclusion

OpenAI's move confirms the direction the industry has been heading: tiered models to avoid overpaying, and agents that do instead of answer. For a small or medium B2B team, the immediate win is migrating to Terra (same results, half the cost) and testing ChatGPT Work on a specific process before scaling.

At MAMBO, we build these kinds of workflows for our clients every day — from automated reporting to agents that prepare meetings. If you want to see how to apply it to your operation without burning months on experimentation, let's talk.