Here Comes the Sun

OpenAI Launches GPT-5.6: Sol, Terra, and Luna (OpenAI)

OpenAI launched the GPT-5.6 series: Sol, the flagship, plus Terra for everyday work and Luna as the fast, cheap tier. Sol is priced at $5 input and $30 output per million tokens, exactly half of Fable 5's rates, and ships with two new modes: max reasoning for deeper single-chain thinking and ultra mode, which spawns coordinated subagents inside the model itself.

The sharpest analysis of launch week fits inside one X post. While the benchmark charts were still loading, AI practitioner Peter Gostev posted his hands-on comparison of the two frontier heavyweights. He cast Claude Fable 5 as the thoughtful, well-spoken wise owl of the pair, and GPT-5.6 Sol as "a rottweiler who will grab the problem by the throat and not let go." Sam Altman quote-posted it with four words: "i do love rottweilers." Just like that, launch day had its defining image.

Here's my take. I don't want a rottweiler on my team. For strategy-level work, the thinking IS the job, and I want the owl. I've always said ChatGPT is Claude's bitch, and this launch didn't change my mind. It clarified the org chart. The owl plans. The rottweiler executes. And a dog that finishes every single task you hand it absolutely has a place in the kennel.

What to do about it: Audition Sol this week, but for a specific seat. Hand it your well-defined execution lists, not your strategy work. The benchmark charts are OpenAI-reported and nobody has published cost per completed task yet, so your own workload is the only test that counts. Full breakdown of what shipped and what early testers are saying is in my new post: Read it here.

📌 Also Worth Knowing

GPT-5.6 Sol Review: Faster Coding, Half Fable 5 Cost, and a Benchmark Problem (Tech Times) — Independent evaluator METR found Sol gamed its agentic benchmark at the highest rate ever recorded, so treat every launch-week chart as directional and run your own workload tests before moving a dollar. searchenginejournal

OpenAI Hiring Points To Image & Video Ads Coming To ChatGPT (Search Engine Journal) — New image, video, and conversational ad formats would give paid placements a bigger footprint in ChatGPT answers, which raises the stakes on the organic AI visibility work you should already be doing. marktechpost

OpenAI Releases GPT-Realtime-2.1 and 2.1-mini for Low-Latency Voice Agents (MarkTechPost) — Reasoning just reached the cheap voice tier at no price increase, which matters if a voice agent for customer questions or lead capture is anywhere on your roadmap. openai

📊 The Number That Made Me Spill My Green Tea

10x

That's roughly how much more often GPT-5.6 goes beyond user intent compared to its predecessor, per OpenAI's own system card: severity-3 behaviors, the kind a reasonable user would strongly object to, showed up in about 0.251% of internal coding-agent traffic. The absolute rate is low, but the direction is the story. A model that interprets instructions too permissively in code may be the same model that's hard to steer in a draft. When you test Sol, watch what it does that you didn't ask for. (Source: my full breakdown)

🧪 One Tool I'm Testing

• ChatGPT Work

• Function: OpenAI's new agentic layer powered by GPT-5.6 that pulls context from your team's tools and files, plans the approach, and produces finished spreadsheets, docs, and slides.

• Marketing application: Turning scattered campaign notes into share-ready decks, competitive journey benchmarking (Virgin Atlantic used it to compress a weeks-long competitor analysis cycle into hours), and recurring reporting workflows you build once and rerun. This is OpenAI's answer to Claude Cowork, so the head-to-head writes itself. lisapeyton

• Pricing: Not listed on the product page. Available on macOS desktop now, rolling out to Windows, web, and mobile across Plus, Pro, Business, Enterprise, and Edu plans. lisapeyton

🏆 My Take: What This Means for Your Marketing

OpenAI just shipped a strong executor at half Fable's sticker price, wrapped in benchmarks nobody can independently verify yet. The routing lesson is the same one you use with your human team: strategy and voice-sensitive work stay with the owl, well-defined execution lists go to the dog. Worth a look, not a migration.

✍️ From Me This Week

A Wise Owl and a Rottweiler Walk Into Your Content Marketing Workflow — My full first-impressions breakdown of GPT-5.6 Sol vs Claude Fable 5 for content marketers: who's saying what (with affiliations attached), the writing verdict split, and the three fine-print caveats to check before you move a workflow. Read it here.

Before You Go

Hit reply and tell me: what's your biggest AI marketing challenge right now?

I LOVE questions, and it helps shape future newsletters.

That's a wrap on this week's AI marketing alchemy. Keep experimenting, keep transforming, and remember: true magic happens when we blend AI efficiency with human creativity.

Crafted with passion and a dash of AI alchemy by Lisa Peyton
AI Practitioner, Professor, and Pioneer

Let's connect: https://linktr.ee/lisapeyton

AI Visibility for Marketing and Comms: Test What AI Says About Your Brand
New 3-hour workshop, August 5th, virtual

You don't need a tracking tool to find out where you stand in AI search. You already have what it takes to run the test yourself. In this workshop you'll run it across ChatGPT, Claude, Perplexity, and Gemini and see exactly what the engines say about your brand. This is the smart first move, the strategic step that comes before tool selection. You walk out with a clear go/no-go verdict you can take to your leadership and defend with evidence, before you spend a single dime on software. Come curious. Leave as the person in the room who knows exactly where the brand stands and what to do next.
Founding rate is live. Code FOUNDERS takes $150 off → https://maven.com/lisa-peyton/ai-visibility-for-communications-workshop?promoCode=FOUNDERS