OpenAI's Half-Price AI Model: What It Really Means for Small Business Automation

A cheaper AI model is not exciting by itself. What's exciting is what becomes affordable once a model gets cheap enough to run thousands of times a day without blinking. That's exactly what OpenAI's latest price cuts do, and the real story isn't enterprise automation, it's that the same math now works for a 20-person business in Pune or Surat.

PYMNTS reported on a lower-cost OpenAI model aimed squarely at enterprise automation, referring to it as GPT-6.1 Terra. Here's where it gets interesting for anyone trying to verify this for themselves: OpenAI's own published announcement doesn't mention a model called GPT-6.1 Terra at all. It talks about two models, GPT-6 Sol and GPT-6 Luna, which received straightforward 50% API price reductions compared to their GPT-5.6 promotional pricing.

This mismatch matters, not because it changes the underlying story, but because it's a useful lesson. AI model names and version numbers move fast (OpenAI alone has shipped GPT-5.6, GPT-6 Sol, GPT-6 Luna and GPT-6.1 Sol within weeks of each other), and headlines sometimes get a model name wrong or conflate two announcements. If you're making a buying decision based on a news headline, always check OpenAI's own pricing page before you commit budget.

The numbers are concrete and public. GPT-6 Sol now costs $2 per million input tokens and $10 per million output tokens, down from $4 and $20. GPT-6 Luna, the smaller and faster model, costs $0.10 per million input tokens and $0.50 per million output tokens, down from $0.20 and $1.00.

To put that in plain terms: a token is roughly three-quarters of a word. A customer support reply of around 150 words uses roughly 200 tokens. At GPT-6 Luna's new pricing, generating a thousand such replies costs well under a rupee in raw model fees. That's the shift that makes automation at scale genuinely cheap, not just theoretically cheap.

Separately, at its September 29, 2026 DevDay, OpenAI described GPT-6.1 Sol as offering near-flagship intelligence at one-fifth of standard token prices, available to API, Plus, Pro, Business, Enterprise and Edu users. And on October 21, 2026, Google Cloud announced that GPT-6 Astra, GPT-6.1 Sol and GPT-6 Luna were entering its model garden in private preview, letting enterprise customers pick models by performance, latency and cost without separate vendor contracts. The direction across the industry is unmistakable: AI providers are racing each other on price, which is good news if you're the one paying the bill.

OpenAI evaluated GPT-6 Sol on something called AutomationBench 1.0.6, a benchmark that tests end-to-end workflows using 47 different tools across sales, marketing, operations, support, finance and HR. These are exactly the functions a typical small or mid-sized business runs every day: following up with leads, scheduling, raising invoices, answering repetitive support questions, onboarding new hires. OpenAI reports that GPT-6 Sol, at its highest effort setting, outperformed Claude Opus 5 at its maximum effort setting while costing just 9% as much per task.

That 9% figure is the real headline, more than the 50% price cut. It suggests that for well-defined, repetitive business tasks, you no longer need the most expensive flagship model to get reliable results. You need the right workflow design around a mid-tier model, which is a very different (and far more affordable) problem to solve.

The cost of running an automated workflow was never just the AI model's price. It's everything you build around it, the triggers, the data connections, the fallback when the AI gets something wrong.

This is the gap worth closing. PYMNTS framed this as enterprise news, and OpenAI's own Business plan pricing (around $20 per user per month billed annually, or $25 billed monthly, starting at two users, with a premium seat at $100 or $125) is clearly priced for companies with teams and IT budgets. But the AutomationBench categories, sales, marketing, support, finance, HR, describe the daily grind of a distributor, a clinic chain, a D2C brand or a real estate agency just as much as they describe a Fortune 500 company.

What makes automation genuinely affordable for a smaller business isn't the per-seat subscription, it's the per-token API cost, because that's what actually scales with volume. A business sending 50,000 automated WhatsApp or email follow-ups a month cares far more about the $0.10/$0.50 per million token rate on GPT-6 Luna than about any seat price. That's the detail most coverage of this story misses: cheaper tokens lower the marginal cost of automation, and marginal cost is exactly what decides whether a small business can afford to run something at scale.

Cheaper model pricing removes one barrier. It doesn't remove the others. In practice, when a business tries to wire up an AI workflow itself, three things tend to go wrong:

None of this is about the AI model being too expensive or too weak. GPT-6 Sol clearing AutomationBench at 9% of a competitor's cost proves the model side is solved for most common tasks. What decides whether automation actually saves a business money is the engineering around it: how the workflow is designed, tested, connected to real business systems, and kept honest over time.

This is precisely where ODIV's AI workflow automation service comes in. Cheaper model pricing from OpenAI is good news, but it's not a product you can hand to a non-technical team and expect results from on day one. ODIV's engineers build the actual automation: mapping out which parts of your sales, support, finance or HR process can run on AI, connecting it properly to your CRM, spreadsheets, WhatsApp, email or billing tools, and setting up the guardrails so the AI hands off to a human when it should.

ODIV's team works directly inside modern AI build tools like Lovable and Claude Code, alongside conventional engineering practice. That combination is the actual value: the AI tools get you to a working build fast, and experienced engineers make sure it's correct, secure, properly integrated with your existing systems, and maintainable after launch rather than something that breaks the first time your data changes shape. Commissioning this kind of build from ODIV lands in a fraction of the time and cost of a traditional, fully hand-coded custom development project, precisely because the team isn't billing for every line of code by hand.

Concretely, for a business reading this story and wondering what to actually do: ODIV would start by mapping your highest-volume repetitive task, lead follow-up, invoice generation, appointment reminders, support FAQs, estimate how many tasks a month it really involves, and then build an automation around the most cost-effective model for that specific job, with proper fallbacks and monitoring built in from day one. If part of that workflow involves talking to customers, ODIV Engage, ODIV's own WhatsApp Business Platform tool, can handle the messaging side with its Automations Builder and AI Agents.

If you're looking at this news and thinking your business generates enough repetitive work to be worth automating, that's the right instinct. The model costs are no longer the obstacle. Start a chat with ODIV on WhatsApp and we'll walk through what a properly built automation would actually look like for your workflow, and what it would realistically cost to build.