Two frontier models in one week. Grok 4.5 landed Tuesday; on Thursday, OpenAI made its GPT-5.6 family generally available, ending a two-week government-review holding pattern. But here is the twist that matters for marketers: the more important launch this week may not be the model at all. It is what shipped alongside it, ChatGPT Work.

GPT-5.6 arrives in three tiers named for the sun, earth, and moon, Sol (the flagship), Terra (balanced, roughly GPT-5.5-class at half the cost), and Luna (fastest and cheapest). That tiering is genuinely useful, and we will get to it. But ChatGPT Work is the headline for your day-to-day. It is an agentic workspace where ChatGPT takes on multi-hour projects, building spreadsheets, decks, docs, and even small web apps from your connected files, apps, and business tools. As one guide put it, ChatGPT Work "quietly matters more for most people's day-to-day than the model itself." Pair it with Anthropic's Claude Cowork from last week, and the picture is clear: the two biggest AI labs just made "delegate a whole project to an agent" the main event.

In today's newsletter, you'll discover how to put agentic workspaces and tiered models to work in your marketing.

📍 Quick Preview:

Let's get into it.

ChatGPT Work Arrives: From Chatbot to Project-Finishing Coworker

On July 9, OpenAI moved its GPT-5.6 family (Sol, Terra, and Luna) to general availability across ChatGPT, Codex, and the API, its first appearance in ChatGPT at all, after a government safety review tied to the model's cybersecurity capabilities cleared faster than expected. But it launched jointly with ChatGPT Work, an agent mode where ChatGPT takes on multi-hour projects, turning information from your files, applications, and connected business services into completed deliverables like spreadsheets, decks, docs, and small web apps.

Here is why that reframes things for marketers. A chatbot answers a question; an agentic workspace completes a project. The shift is from "help me draft this" to "build me the thing." Feed ChatGPT Work your connected tools and a clear brief, and it can research, assemble, and produce a finished first draft of the deliverable, campaign brief, competitive analysis deck, performance report, landing-page mockup, while you supervise. It is the same "delegate and review" model we saw with Claude Cowork last week, now from the other giant, which tells you this is the direction of the whole industry, not one company's bet.

The model tiering underneath it is the practical money story. OpenAI is explicitly framing GPT-5.6 as a routing decision, not a single model: use Luna for high-volume simple work (drafts, classification, extraction), Terra for everyday production work at roughly half the cost of the prior generation, and Sol for heavy reasoning where it counts. For marketers watching AI spend, that means you no longer default to the most expensive model for everything, you match the tier to the task. (Two honest notes: Sol access requires eligible paid plans, and independent coverage stresses GA day is "an access story, not a benchmark story", so test on your real work rather than trusting headline evals.)

1. GPT-5.6's Tiers Make "Match the Model to the Task" Official
OpenAI is pitching GPT-5.6 as a routing family: Luna for cheap high-volume work, Terra for balanced production (GPT-5.5-class at about half the cost), Sol for frontier reasoning, with Sol priced at $5/$30, Terra $2.50/$15, and Luna $1/$6 per million tokens.
Why This Matters: The cost-conscious play we have advocated all month is now baked into the product. Defaulting to the flagship for everything is the expensive mistake.
✅ Action Item: Map your AI tasks to tiers, route drafts and classification to the cheapest model, reserve the flagship for high-stakes reasoning, and watch your costs drop.

2. The Two-Week Model Gate Closes, With a Lasting Lesson
GPT-5.6's release followed a US government safety review (via Commerce's Center for AI Standards and Innovation) tied to its cyber capabilities, the same dynamic that briefly pulled Anthropic's models in June.
Why This Matters: Model availability is now subject to review and can shift on policy timelines, reinforcing why flexibility across providers beats locking into one.
✅ Action Item: Keep a second capable model viable in your stack so a review, delay, or restriction never stalls a critical workflow.

3. Codex Folds Into the ChatGPT Desktop App
OpenAI merged Codex into a new ChatGPT desktop app for Mac and Windows, so Chat, Work, and Codex live behind one front door rather than separate apps.
Why This Matters: The trend is consolidation, fewer app islands, more unified workspaces, which lowers friction for teams adopting AI into daily work.
✅ Action Item: Favor unified AI workspaces over scattered single-purpose apps; adoption and consistency both improve when everything lives in one place.

4. "GA Is an Access Story, Not a Benchmark Story"
Independent analysis stressed that GPT-5.6's general availability is mostly about access and new API features (max reasoning, an ultra multi-agent mode, programmatic tool calling), not a benchmark leap, and noted Claude models still lead several of OpenAI's own comparison rows.
Why This Matters: Launch-day hype rarely equals real-world superiority. The honest read is "capable and now accessible," not "new undisputed champion."
✅ Action Item: Ignore the leaderboard drama; run five of your real tasks side by side and pick the cheapest tier that clears your quality bar.

5. ChatGPT Work vs. Claude Cowork: The Agentic Workspace Race Is On
With ChatGPT Work launching a week after Claude Cowork went cross-device, the two leading labs are now competing head-to-head on "delegate a whole project to an agent."
Why This Matters: Direct competition means faster improvement and better pricing on exactly the capability most useful to lean marketing teams.
✅ Action Item: Try both on the same real task and see which fits your stack and workflow, then standardize on the winner for your team.

6. Ultra "Multi-Agent" Modes Push Parallel Work (at a Cost)
GPT-5.6's new ultra setting coordinates multiple agents in parallel by default for complex, multi-stage work, finishing faster but consuming more tokens.
Why This Matters: Parallel agent workflows can compress big projects dramatically, but the token cost means you reserve them for genuinely complex jobs.
✅ Action Item: Save multi-agent "ultra" modes for your most complex projects; use single, cheaper models for routine tasks to control spend.

7. Faster Inference Hardware Enters the Picture
OpenAI said it is bringing GPT-5.6 Sol to specialized hardware at up to 750 tokens per second for select customers, part of the broader race to make frontier AI faster and cheaper to serve.
Why This Matters: Faster, cheaper inference keeps flowing downstream into the tools you use, continuing the cost-down trend that benefits every marketer.
✅ Action Item: Expect your AI tools to get faster and cheaper; re-evaluate usage-based costs at renewal to capture the savings.

8. The Real Bottleneck Is Still Skill, Not Access
Coverage of the launch landed on a familiar note: "the upgrade came to you; the skill of using it well still doesn't", echoing the enablement gap we covered with HubSpot's guide.
Why This Matters: A more powerful model does nothing for a team that has not learned to delegate to and direct it well. Fluency remains the differentiator.
✅ Action Item: Pair every model upgrade with a quick team enablement step, one shared example of a great agentic-workspace brief goes a long way.

ChatGPT Work - Turn Your Connected Tools Into Finished Deliverables

This week's real story is the agentic workspace, so the spotlight goes to ChatGPT Work itself, the surface OpenAI clearly considers the future. Unlike a standard chatbot, ChatGPT Work operates as an agent that takes on multi-hour projects: it pulls from your connected files, apps, and business services, then produces completed deliverables, spreadsheets, decks, documents, even small web apps. For marketers, it is the difference between an assistant that helps you write and a coworker that builds the thing.

🔍 DEEP DIVE: From Answers to Deliverables

What's Happening:
Agentic workspaces move AI from conversation to completion. You connect your data and tools, hand over a scoped project, and the agent researches, assembles, and produces a finished first draft, surfacing decisions to you along the way. ChatGPT Work is OpenAI's entry, powered by the GPT-5.6 family underneath.

Why It Matters:
The deliverables that consume a marketer's week, reports, decks, analyses, briefs, are exactly what these agents are built to produce. Offloading the assembly frees your time for strategy, judgment, and the creative work that actually differentiates.

Key Implications:

Looking Ahead:
Expect agentic workspaces (ChatGPT Work, Claude Cowork, and more) to become a primary way marketers get work done, with the model humming underneath and the human directing.

What You Can Do With It:

Pricing: Included with eligible ChatGPT plans (tiers vary; free users get lower-cost models) - confirm current terms
Official Website: ChatGPT

A Full AI Ad Creative Studio (Just Paste Your URL & Get AI Ads in Seconds)

If you've been following the AI Marketing Newsletter, you know how powerful the right creative can be. And you know we love leveraging the power of AI to amplify results!

Now imagine having AI that reads your website - your colors, your tone, your imagery - and turns it into scroll-stopping ad creative across every format in seconds, not sprints.

Key Features of AdsCreator.com

As you'd expect from someone who's generated 150M+ YouTube views & built multiple AI-powered marketing tools, we've built AdsCreator.com with practical, results-oriented features:

1. URL-to-Ad Generation Paste any website URL, and get polished, campaign-ready ad creative — no briefs, no back-and-forth.

2. Brand DNA Extraction Our AI scans your site's colors, typography, tone of voice & product imagery — so every ad feels on-brand from the jump.

3. Every Format in One Pass Meta, Google, Stories — square, landscape & vertical formats generated from the same creative direction instead of rebuilt one by one.

4. Zero Design Skills Required If you can paste a URL, you can create professional ad creative in 60 seconds. No Canva. No retainers. No waiting.

5. A/B Testing at Scale Generate multiple ad variations instantly to test angles, hooks & creative concepts — without starting from scratch every time.

6. Ship Creative Faster Go from blank page to testable ad creative in minutes, not days. More tests. More data. More winners.

Notion AI - Connected Workspace With Agents

Gamma - AI Deck & Document Generation

Hex - AI-Powered Data Workspace

Guru - AI Enterprise Knowledge & Answers

This AI Marketing Newsletter is Sponsored by:

Sign up for a free trial of Attio, the agentic CRM.

Ask Attio to build a daily workflow that surfaces the deals that need your attention today, like anything with a stage change, a recent reply, or a new signal in the last 24 hours.

Review your pipeline in Claude, synced live from Attio via MCP.

That's it.

Try Attio Now

Build a "Tier + Delegate" System to Get More From Agentic AI at Lower Cost

With tiered models and agentic workspaces now standard, the highest-leverage move is a simple system that captures both the capability and the savings. Here is the playbook:

Success Metric: Track two numbers, cost per completed task (should fall as you route work to cheaper tiers) and hours reclaimed per delegated deliverable (should rise as your briefs improve). When both move in the right direction, you have a system that captures the capability of agentic AI and the savings of tiered models, exactly the combination this week's launches made possible.

Here is the thread tying today together: agentic AI is powerful, but it delivers most when you point it at the right job with the right inputs. The best agentic workspace still needs sharp, high-intent targeting for the campaigns it helps you build, or all that produced creative reaches the wrong people.

That is where KeywordSearch.com's AI Audience Builder fits. Instead of hours of manual audience research, you generate high-intent, AI-powered audiences in seconds and sync them straight to Google and YouTube Ads with one click. As agents take over more of the building, precise targeting is what ensures everything they produce actually lands and converts.

Supercharge your ad targeting with AI-powered audience creation.

→ Start your unlimited 5-day free trial and build your first AI-powered ad audience in about 2.6 minutes.

We will be back tomorrow with more.

Aleric & Greg KeywordSearch.com

P.S. As agents build more of your marketing, keep the targeting sharp so it all lands. Start your unlimited 5-day free trial of KeywordSearch and build your first high-intent audience in minutes, the precise input that makes everything your AI produces actually convert.

This AI Marketing Newsletter is Also Sponsored by:

Most finance teams work on five different tools and a prayer. Ramp replaces all of it: corporate cards, bill pay, expense management, travel, and procurement. AI-powered with real-time visibility and real control.

Sign up for Ramp today