OpenAI announced unlimited ChatGPT text chats for free users, powered by GPT-5.6 Luna, rolling out the week of August 10. Here is what service businesses need to know and do.
Ido Cohen · Published 2026-08-09 · AI for Service Business
OpenAI dropped a significant free-tier upgrade on August 6, 2026: GPT-5.6 Luna is now the default model for every free ChatGPT account, and the old 10-messages-per-5-hours cap is gone — unlimited text chats rolling out starting the week of August 10. If you run a plumbing company, law firm, dental practice, or HVAC business, this matters because it changes what your customers expect, what your employees can do with zero budget, and what your competitors can pull off without a paid subscription.
---
This is not a minor version bump. According to TechCrunch's August 6 coverage, OpenAI is "removing limits on text-based chats for all users on ChatGPT, which recently crossed the 1 billion weekly user mark." That scale is the headline buried inside the upgrade.
Here is the specific breakdown of what changed by tier:
The key clarifications, because the internet has gotten this wrong:
The honest take: GPT-5.6 Luna is a capable everyday model. OpenAI's internal evaluation claims it produces 62–68% fewer factual errors than GPT-5.5, though as implicator.ai noted, those figures had not been independently verified as of August 7. For drafting emails, answering customer questions, and summarizing notes, it is more than adequate. For complex legal or medical advice generation, the limitations matter.
---
One billion weekly users now have unlimited access to a frontier-class AI — and most of them are your customers, your staff, and your competitors' staff.
Think about what unlimited free access actually enables:
The price war among AI providers is making this worse (or better, depending on how you look at it). According to MarketingProfs' August 7 AI Update, DeepSeek released V4 Flash the same week, "a coding model that approaches the performance of Anthropic's Claude Opus 4.8 while costing roughly 99% less for comparable output." The AI price war now includes OpenAI, Google, Meta, and DeepSeek all racing to give users more for less. The cost of AI capability is trending toward zero for basic tasks.
For a service business owner, the strategic read is simple: the AI playing field just leveled even further. Your customers have it. Your employees have it. The question is whether you are ahead of them or behind them.
---
Luna is optimized for low latency and cost efficiency — it is OpenAI's "most cost-efficient model," as Unite.AI described it, designed for high-volume everyday work. Here is a realistic skills assessment for a service business use case:
Luna handles well:
Luna struggles with:
For most day-to-day marketing and operations tasks at a service business, Luna is fine. You do not need to be on a paid plan to get real value out of ChatGPT in August 2026. But if your team is doing anything mission-critical — generating legal language, building automations, analyzing data files — Plus at $20/month is still the right call.
---
This is the part most of the coverage glosses over, and it is the part that could hurt you.
Free ChatGPT users' conversations feed OpenAI's training data by default. According to gagadget's reporting, you need to opt out manually under Settings → Data Controls. If your employees are using free ChatGPT to draft client proposals, paste in patient information, discuss a pending legal case, or write estimates with specific customer data — that content is potentially going into OpenAI's training pipeline.
Three rules to implement today:
1. Never paste identifying client information into free ChatGPT. Names, addresses, case details, medical history, financial figures — none of it.
2. If you want your team using ChatGPT for business work, pay for Plus ($20/month) or Business ($20/seat). Paid tiers do not train on your data, per OpenAI's terms.
3. Set a clear internal policy now. The unlimited free tier will dramatically increase casual AI use among your staff. Without a policy, you cannot control what gets pasted in.
This is not hypothetical risk-management theater. Service businesses handle sensitive information — health records, financial situations, legal matters, home addresses — constantly. The unlimited free tier is convenient, but "convenient" and "compliant" are different things.
---
Here is the competitive reality check: the service businesses in your market that were already using ChatGPT occasionally were rationing their messages. That friction is now gone. Expect to see more AI-generated content, faster response times, and more polished proposals from competitors who were previously limited by the 10-message cap.
The Advantage+ parallel from Meta's ad platform is instructive here. According to The Next Web's reporting on eMarketer's 2026 forecast, Meta's AI-powered Advantage+ ad automation drove 24.1% growth — nearly double Google's 11.9% — because it removed friction from ad creation. When AI tools remove friction, usage explodes. That is exactly what OpenAI is doing to ChatGPT's free tier.
The businesses that win from this shift are the ones that move first. If you are still drafting every estimate by hand, writing every review response manually, and starting every email from scratch — you are now competing against operators who are not doing any of those things.
---
If you are paying for any AI-powered marketing or CRM tool right now, there is a good chance it runs on GPT-5.6 under the hood. According to eesel.ai's pricing breakdown, OpenAI cut Luna's API prices by 80% and Terra's by 20% on July 30, 2026. That means the AI tools you pay for should be getting cheaper to run, and in competitive markets, some of that savings will flow back to you as a customer.
Quick model guide for service business owners who encounter these names in their tools:
The practical upshot: when you see a tool advertise "GPT-5.6 powered," ask which model. Sol is genuinely excellent. Luna is good but not the same.
---
Here is a concrete action list, ordered by priority:
1. Open ChatGPT's free tier and test Luna yourself. By August 10, the unlimited rollout will be live. Spend 20 minutes throwing real work tasks at it — writing a review response, drafting an estimate email, summarizing your service list. Know what it can do before your staff figures it out on their own.
2. Set an AI data-use policy for your team — this week. Write a one-pager: what can and cannot be pasted into free AI tools. Send it in your team chat. It does not need to be a legal document; it just needs to exist.
3. Audit your current paid AI subscriptions. If you or a team member is paying for a tool primarily to get unlimited ChatGPT text access, that justification is now weaker. Consolidate where possible. But keep any paid plan where your data confidentiality is at stake.
4. Use the Think button for harder tasks. Starting August 10, free users get a Think button that triggers deeper reasoning from Luna. For a complicated response to a negative review or a tricky estimate scenario, hit Think before accepting the first output.
5. Assign one team member to test ChatGPT for one specific workflow. Pick your most time-consuming text task — writing job follow-up emails, drafting Google Business Profile posts, responding to online inquiries. Have them run the AI version for two weeks and compare quality and time saved. One real test beats ten blog posts.
6. Check your opt-out settings. If you or any team member uses free ChatGPT for anything work-related right now, go to Settings → Data Controls and toggle off model training. Do it today, before the volume of use increases next week.
---
Is GPT-5.6 Luna actually a good AI model for a service business?
For everyday writing and communication tasks — drafting emails, responding to reviews, creating social posts, answering common customer questions — yes, Luna is more than capable. It is OpenAI's lowest-tier model in the GPT-5.6 family, optimized for speed and cost efficiency rather than deep reasoning. It will not replace the judgment of your best employee, but it will handle the repetitive writing work that takes up too much of their time.
Does the unlimited free tier mean I can cancel my ChatGPT Plus subscription?
Not if you are using ChatGPT for anything involving client data, file analysis, or complex decision-making. The paid Plus plan ($20/month) gives you the more capable Sol model, removes ads, unlocks file uploads, and — critically — does not train on your conversations. For a service business owner with sensitive client information, the $20/month is worth it. For a team member who mostly uses ChatGPT to draft text, the free tier may now be sufficient.
Will my customers be using this more because it is unlimited now?
Almost certainly. ChatGPT already had 1 billion weekly users before the unlimited rollout. Removing the message cap will increase casual daily use significantly. Practically, this means more customers arriving at your business better-informed about their problem — and sometimes with incorrect AI-generated information they believe is correct. Train your front-line staff to ask what a customer already knows and to correct misconceptions diplomatically.
What is the difference between Luna's Think button and just asking a harder question?
The Think button tells Luna to slow down and work through a problem step by step before answering, rather than giving the fastest response. It is useful for multi-part problems, complex instructions, or situations where the first answer has been wrong. It does not switch you to a stronger model like Sol — you stay on Luna. For truly complex work, Sol (paid) is still the better tool.
Should I be worried about my employees using free ChatGPT for work tasks?
Yes, with one specific concern: data privacy. Free ChatGPT conversations can be used to train OpenAI's future models by default. If your team members paste in client names, addresses, financial details, or health information, that data leaves your control. The fix is simple: set a policy that work-related AI use happens on paid accounts (which do not train on data), and that identifying client information is never pasted into any AI tool. Get that policy in writing before the unlimited rollout goes live next week.
---
Sources: