AI This Week: Price Cuts, Meta's SMB Push, Google's New Flash
Three moves worth flagging this week: OpenAI’s flagship model got meaningfully cheaper for the third time in a month, Meta opened real business data integrations to SMB owners at no cost, and Google shipped a new Flash model priced at half its predecessor. The AI pricing floor keeps dropping, and new tools keep entering reach for businesses that weren’t in the market a year ago.
OpenAI Cut GPT-5.6 Sol Prices by Up to 33% Through November
On August 21, Reuters reported that OpenAI quietly reduced pricing on its flagship GPT-5.6 Sol model — its most capable and previously its most expensive. The new rates drop input tokens from $5 to $4 per million (down 20%) and output tokens from $30 to $20 per million (down 33%). Cached input also fell from $0.50 to $0.40 per million. OpenAI posted the rate as promotional, running through at least November 21, 2026. The cuts apply to pay-as-you-go API access, Codex credits, and eligible ChatGPT Work plans — not consumer Pro, Plus, or Business subscriptions.
This is the third price reduction on the GPT-5.6 model family in under a month. Per Reuters’ coverage, Sol now costs less per token than Anthropic’s Claude Opus 5 on both input and output — a notable reversal for a model that held its price for months while competitors cut theirs. The broader context: Google’s Gemini 3.7 Flash (covered below) launched at roughly half the previous generation’s price the same week, and Anthropic has continued chipping away at enterprise accounts.
The practical read: these cuts are on the API tier, meaning they affect businesses paying directly for access through custom integrations, not businesses using consumer subscription plans. But they signal the direction of travel — frontier-model costs have dropped dramatically over the past year, and the trend isn’t slowing.
What this means for your business: If your team is building AI into operations at the API level — automating data entry, building custom assistants, processing documents at scale — GPT-5.6 Sol just got materially cheaper through the end of November. Freight and logistics and wholesale distribution operations running high-volume document or dispatch workflows are the ones most likely to see a meaningful cost difference. If you’re on a consumer subscription, nothing changes for now.
Meta Opened Its Business Data to AI — Free for SMB Owners
On August 19, Social Media Today reported that Meta rolled out a new integration layer for its AI assistant, specifically aimed at small and mid-size business owners. The update lets owners connect Meta AI to their Facebook and Instagram accounts, Meta ad campaigns, and Google Workspace — meaning Gmail, Docs, Sheets, and Slides. Once connected, Meta AI can analyze ad performance, surface insights on what’s working, and turn findings into drafts, decks, and spreadsheets through conversational prompts.
The new integration features are available for free across meta.ai, the Meta AI mobile app, and the newly launched Meta AI desktop app for Mac. This builds on Meta Business Agent, which Meta originally announced at its Conversations conference in London in June and has been expanding since. Business Agent handles customer-facing interactions — answering questions, booking appointments, qualifying leads — across WhatsApp, Instagram, and Messenger. The August 19 update is specifically about the owner-facing analytics and workflow side.
One important note: Meta Business Agent itself moved to token-based billing on August 1, at $2 per million tokens. The new analytics integrations announced this week are free, but the broader customer-interaction automation layer is now a paid product.
What this means for your business: If you’re already spending money on Meta ads — Facebook, Instagram, or both — this integration is worth testing. Connecting Meta AI to your ad account and a Google Sheet takes minutes, and being able to ask conversational questions about what your campaigns are doing is genuinely useful for owners who don’t have a dedicated marketing analyst. Businesses in commercial real estate, construction, and specialty contracting that run local lead-generation ads on Meta should look at this specifically — the insights layer is free, and understanding where your ad spend is going has real dollar value.
Google’s Gemini 3.7 Flash Is Google’s Best Workhorse Model at Half the Old Price
On August 13, Google announced Gemini 3.7 Flash on its official blog, calling it its “most intelligent workhorse model” yet. The key spec: it launched at half the per-token cost of Gemini 3.6 Flash — the previous generation that itself only launched three weeks earlier. The introductory rate is $0.75 per million input tokens and $3.75 per million output tokens, with those rates set to increase on January 1, 2027. Google also cut Gemini 3.6 Flash’s price to match on the same day.
Per Google’s announcement, Gemini 3.7 Flash is optimized for coding, agent tasks, and software engineering workflows, with strong gains in first-pass code accuracy and production-ready code generation. It’s available via Google AI Pro and Ultra subscribers through the Spark agent interface, and through the Gemini API for developers and businesses building custom integrations.
The pattern here mirrors what OpenAI is doing: the “workhorse” tier — not the reasoning-heavy frontier models, but the fast, capable models used for real automation workloads — keeps getting cheaper and more capable at the same time. That combination is what brings AI into practical reach for businesses that don’t need the most powerful model for every task.
What this means for your business: If your team or vendors are using AI APIs to automate tasks — processing invoices, summarizing emails, drafting reports — the cost per task just dropped again. Energy services and multifamily property managers running AI-assisted document workflows or tenant communications are the types of operations where per-token cost shows up in the monthly bill. And if you’re building something new, this is the tier worth starting with: capable, fast, and significantly cheaper than frontier models for most business tasks.
The Takeaway
This week’s story is price compression. OpenAI’s flagship model is 33% cheaper than it was a week ago. Google’s latest workhorse model costs half what its predecessor did three weeks ago. Meta opened free business analytics integrations to any owner with an ad account and a Google Workspace login.
The tools available to a Houston small business owner in August 2026 cost a fraction of what the same capabilities cost twelve months ago — and the trend isn’t slowing. The businesses that benefit most aren’t the ones waiting for costs to hit zero. They’re the ones figuring out where AI actually fits their operations now, while competitors are still on the sideline. If you want help mapping what any of these tools mean for your specific workflow, BlueHill is here.