New models arrive faster than a marketing team can test them, so the useful question is rarely which one is best. It is which task you would give each one, what that costs per finished deliverable, and how you would notice if a cheaper model were good enough.
These posts are working notes from using the models on real marketing jobs: briefs, drafts, audits, reporting and agents. Prices and version names are dated in each post, because both go stale within weeks.
104 articles2 guides and pages
Start here
A short reading order, if you would rather not browse.
- 01Which AI Model for Marketing? A Release-Fatigue FrameworkA framework for choosing without chasing releases.
- 02How to Evaluate AI Models for Marketing: A Test ProtocolA repeatable test protocol.
- 03AI Model Pricing Compared: The Marketer's Version 2026The cost view a marketer needs.
- 04The Model Routing Strategy for Marketing TeamsCheap model by default, expensive model on purpose.
- 05Marketing Tasks Worth Handing to an Agent (and Which Aren't)What to hand to an agent and what to keep.
How I read a model launch
I wait for a task I actually do, run it through the old and new model, and compare edit time and cost. A launch post is not evidence. The evaluation protocol and routing posts below are the method; the comparison posts are applications of it that will age out.
Where this meets client work
Two places the workflow shows up in shipped projects.
Claude and Claude Code 24
Models, effort levels and building with Claude Code.
- Claude Code Skills for SEO Audits: A Reusable Workflow14 Oct 2026Build a reusable SEO audit skill in Claude Code: SKILL.md structure, a worked template, safe tool permissions and where human review still matters.
- Gemini 4 vs GPT-6.1 vs Claude Opus 5.5: Marketer's Guide10 Oct 2026Gemini 4 Argon, GPT-6.1 Sol and Claude Opus 5.5 compared on price, access, context and benchmarks for marketers, plus a bake-off test. As of 14 Oct 2026.
- Anthropic Cyber Mission and OSS Scanner: What Marketers Need9 Oct 2026Anthropic's Cyber Mission (8 Oct 2026) adds a free OSS Scanner and an 11-partner infrastructure program. Here is what it means for your site and stack.
- GPT-6.1 Sol vs Claude Opus 5.5: Alignment Index Scores9 Oct 2026Arena's Alignment Index puts GPT-6.1 Sol at 87.9 and Claude Opus 5.5 at 83.2. See what it measures, where it falls short, and what marketers should do with it.
- Claude Haiku 5.5 vs Sonnet 5.5: Cheap-Model Content Ops9 Oct 2026Claude Haiku 5.5 costs $0.10/$0.50 per million tokens vs Sonnet 5.5 at $2/$10. Here is which content tasks to route to each, with cost math.
- Claude Code 2.1.28x: The Updates Solo Marketers Should Use27 Sept 2026Claude Code shipped versions 2.1.280 to 2.1.285 in late September 2026. Here are the changes solo marketers can use, and the ones you can safely skip.
- When Claude Goes Down: A Marketing Team Outage Playbook27 Sept 2026Claude had a multi-surface outage on 29 September 2026. Here is what was reported, and a practical fallback playbook so your content team never stalls.
- Claude Plugin Directory Portal: How Marketers Can Submit27 Sept 2026Anthropic opened a plugin submission portal on 25 Sept 2026. Here is how marketers can package a plugin, pass the automated scan, publish it and track installs.
- Claude Sonnet 5.5 for Marketing: Use Cases and Cost Math27 Sept 2026Anthropic launched Claude Sonnet 5.5 on 28 Sept 2026 at $2/$10 per million tokens. Marketing use cases, worked costs and limits to know before switching.
- Perplexity vs Claude Code: What Marketers Should Learn Here27 Sept 2026Perplexity Computer and Claude Code are often compared, but they do different jobs. Here is what marketers can learn about choosing agents in Sept 2026.
- Sonnet 5.5 vs Opus 5.5 vs Fable 5.1: Which for Content27 Sept 2026Sonnet 5.5, Opus 5.5 and Fable 5.1 differ a lot in price. Here is which Claude model to use for which content job, with costs, tests and caveats.
- Claude Fable 5.1 Effort Levels for Marketing Writing Guide18 Sept 2026How to pick Claude Fable 5.1 effort levels (low to max) for blogs, ads, emails and strategy docs, with a task map, cost math and workflows as of September 2026.
- Claude Fable 5.1 for Marketing Teams: The Real Story12 Sept 2026Claude Fable 5.1 launched 1 September 2026. The headline is the model, but the 75% cache-read price cut is what actually changes marketing team economics.
- Fable vs Mythos: Why You Can't Use One of Them12 Sept 2026Claude Fable vs Mythos: the same underlying model with different safeguard levels. Fable is generally available. Mythos is gated to vetted professionals only.
- Claude's Text Watermark: What Marketing Teams Need to Know12 Sept 2026Anthropic detailed Claude's text watermarking on 14 August 2026. Here's what AI text watermarking means for marketing teams publishing AI-assisted content.
- GPT-6 Astra vs Claude Fable 5.1 for Marketing Work12 Sept 2026GPT-6 Astra vs Claude Fable: identical $10/$50 pricing, different caching economics. A practitioner comparison for real marketing workflows, September 2026.
- Claude Citations May Be a Brave Search Backdoor28 Aug 2026Claude's web search is Brave-backed. A small study found 86.7% overlap between Claude citations and Brave organic results. Here's the thesis and how to test it.
- The Claude + Claude Code Marketing Stack for 202625 May 2026What a real Claude marketing agency stack looks like in 2026, Claude for strategy and writing, Claude Code for landing pages and automations.
- Claude AI Prompts for Marketing: Edtech & Agency Workflows7 May 2026Practical Claude AI prompts and workflows for edtech and agency marketing teams, persona research, admissions copy, audits, and more.
- Claude vs ChatGPT for Marketing Content: A Practical View5 May 2026An agency practitioner's honest comparison of Claude vs ChatGPT for marketing content, brand voice, long-form writing, and real use cases.
- Using Claude to Audit Competitors' AI Search Presence3 May 2026A practical workflow for using Claude to audit how competitors show up in AI search, a repeatable quarterly AEO process for solo marketers.
- Claude Code for Marketers: Build a Landing Page, No Dev1 May 2026A step-by-step guide for marketers using Claude Code to build a landing page in plain English, realistic expectations, no coding background needed.
- Claude Code Marketing Automation Case Study (Reporting)29 Apr 2026A real Claude Code marketing automation case study: how I automated one weekly client reporting workflow, cut hours off it, and what still needs a human.
- Claude Code for Programmatic SEO Course Pages (Edtech)27 Apr 2026How to use Claude Code to build city x course landing pages for edtech and bootcamps without creating thin, duplicate programmatic SEO pages.
OpenAI and ChatGPT 27
GPT releases, plans and the Agents API.
- ChatGPT's Visual Interface and Visual Ads: Brand Guide9 Oct 2026ChatGPT's Intelligent UI rolled out 7-8 Oct 2026, and visual ads start testing in the US this month. What brands should watch, test and ignore.
- Gannett (USA Today Co) Sues OpenAI: Licensing Lessons9 Oct 2026USA Today Co, formerly Gannett, sued OpenAI on 8 Oct 2026 seeking over $250M. What is alleged, what is not proven, and what publishers and brands should do.
- Rank in ChatGPT in 7 Days? What Is Actually Possible in 202628 Sept 2026Can you rank in ChatGPT in a week? A fact-checked look at what a fast win looks like, what needs months, and what the citation studies actually show today.
- The Reddit Citation Drop in ChatGPT: Known vs Still Unknown28 Sept 2026Reddit's share of ChatGPT Search citations reportedly fell 86% in August 2026. What Promptwatch measured, the theories, the caveats and the next steps.
- OpenAI Decisions API: Use Cases for Growth Teams27 Sept 2026OpenAI's Decisions API returns fast, confidence-scored choices from a fixed answer list. Growth-team uses, limits and a test plan before pricing lands.
- ChatGPT at 1.2B Weekly Users: What It Means for Traffic26 Sept 2026OpenAI says ChatGPT reaches 1.2B people weekly. What the claim does and does not tell you, and how to check whether ChatGPT sends your site any traffic.
- ChatGPT Dots Agents: What Marketing Work Can You Delegate?26 Sept 2026OpenAI's Dots are always-on ChatGPT agents with their own cloud computers. Here is which marketing tasks to hand over, which to keep, and the guardrails to set.
- ChatGPT Plugin Extensions: A New Brand Discovery Channel?26 Sept 2026ChatGPT plugin extensions add sidebar panels, file viewers and event-driven automations. What it means for brand discovery, and what is still unproven.
- ChatGPT Pro 500 Plan Explained: Who Needs 25x Plus26 Sept 2026OpenAI's $500 ChatGPT Pro plan offers 25x Plus usage and Ultrafast speed. Here is who should pay, who should stay on Plus or Pro 200, and how to test it.
- ChatGPT Team Tasks and Space for Small Agencies: Review26 Sept 2026A workflow review of ChatGPT Space, Pages and Team Tasks for small agencies: what each does, plan requirements, a sample setup and the risks to manage.
- GPT-6.1 Sol: What Changed and Should Content Teams Switch?26 Sept 2026GPT-6.1 Sol costs $2/$10 per million tokens, one-fifth of Astra by OpenAI's account. Here is what changed, the cost math and a fair test before you switch.
- GPT-6 Astra Ultrafast: When 8x Speed Actually Matters (2026)26 Sept 2026OpenAI's GPT-6 Astra Ultrafast promises up to 8x faster output in Codex and 6x in the API. Here is when the speed is worth the premium, and when it is not.
- OpenAI DevDay 2026: Every Announcement Marketers Should Know26 Sept 2026OpenAI DevDay on 29 Sep 2026 brought GPT-6.1 Sol, Dots agents, Team Tasks and plugins. Here is what matters for marketers and what is still unclear today.
- Sign in with ChatGPT: Distribution Play or Lock-In? (2026)26 Sept 2026OpenAI's Sign in with ChatGPT lets subscribers use their plan allowance in 16 launch partners' tools. What it means for brands, SaaS marketers and users.
- SaaS Alternatives Pages SEO When Buyers Ask ChatGPT in 202621 Sept 2026Build SaaS alternatives and comparison pages that rank on Google and get cited when buyers ask ChatGPT for alternatives, with templates and honesty rules.
- Building a Marketing Agent With the OpenAI Agents API Now18 Sept 2026OpenAI Agents API marketing use cases: what launched 10 Sept 2026, six agents worth building, a step-by-step content-refresh agent design, costs and guardrails.
- How to Check if ChatGPT Cites Your Website (Free Methods)15 Sept 2026Free ways to check if ChatGPT cites your website: buyer-intent prompt tests, GA4 referral reports, server logs for OpenAI bots, and a simple tracking sheet.
- Advertising Inside ChatGPT: What You Can Actually Control12 Sept 2026OpenAI's ChatGPT ads serve only to Free and Go tiers across 9 markets. Here's the targeting constraint, the measurement limits, and how to run a first test.
- GPT-6 Astra: What Marketers Actually Need to Know12 Sept 2026GPT-6 Astra launched 3 September 2026 at $10/$50 per million tokens. A practitioner's honest breakdown of pricing, access, and where it fits in marketing.
- How to Migrate to GPT-6 Astra Without Breaking Your Stack12 Sept 2026A practical guide to migrate to GPT-6 Astra from GPT-5.x: enterprise rollout defaults, what to re-test, cost implications at $50/M output, and a staged plan.
- No, OpenAI's Astra Is Not Google's Astra12 Sept 2026OpenAI Astra vs Google Astra: one is a shipped flagship model, the other a research prototype that never launched. Here is the difference, clearly explained.
- ChatGPT Ads Hit Europe: What It Means for Organic AEO28 Aug 2026OpenAI expanded ChatGPT ads to 31 European markets on 24 August 2026. What paid placement inside the answer surface means for organic AEO and visibility.
- Yelp + OpenAI: Local AEO Just Changed for Service Brands28 Aug 2026Yelp licensed its reviews to OpenAI in July 2026. Here's what local AEO looks like when review platforms become retrieval infrastructure for ChatGPT.
- How to Get Cited by ChatGPT and Perplexity: Format Playbook22 Aug 2026How to get cited by ChatGPT and Perplexity: the content formats AI systems actually pull from, why they work, and the off-site corroboration that decides it.
- How to Rank in ChatGPT Search: A Founder's Framework16 Aug 2026Learn how to rank in ChatGPT Search with a step-by-step, low-budget framework for founders: brand mentions, direct-answer content, and citations.
- How Students Use ChatGPT to Research Bootcamps6 Jun 2026How students use ChatGPT and AI tools to research coding bootcamps, compare outcomes, and check legitimacy, and how edtech brands can get found.
- The 'Could ChatGPT Write This?' Originality Audit Test18 Mar 2026A simple pre-publish test and checklist to audit content originality before you hit publish, spot generic content before it costs you rankings.
Gemini and other models 11
Google, open-weight and on-device models.
- Gemini 4 Argon in AI Mode: Who Gets It and the SEO Impact10 Oct 2026Gemini 4 Argon reached AI Mode for Google AI Pro and Ultra subscribers on 13 October 2026. What it is, who has it, and what SEO teams should test this week.
- Gemini 4 Argon API Pricing: Cost Math for Content & Agents10 Oct 2026Gemini 4 Argon costs $2 in and $10 out per million tokens until 30 Nov 2026, then $4 and $20. Worked cost examples for content, bulk jobs and agent runs.
- Open-Weight Models Like StepFun Step 5: Do Marketers Care?10 Oct 2026StepFun's Step 5 Preview has open weights scheduled for 15 October 2026. Here is what open-weight means, what is verified, and when marketers should care.
- Gemini 3.8 Flash in AI Mode: The Citation Bug Explained4 Oct 2026Gemini 3.8 Flash briefly showed almost no source links in AI Mode on 3 September 2026. What happened, how it was fixed, and what marketers should learn.
- Snapdragon 8 Elite Gen 6: On-Device AI and Marketing28 Sept 2026Qualcomm's Snapdragon 8 Elite Gen 6 and Extreme bring on-device AI agents to phones. Here is what is confirmed, what is hype, and what marketers should do.
- Gemini 3.8 Flash in AI Mode: What It Means for SEO Today27 Sept 2026Google added Gemini 3.8 Flash to AI Mode on 2 Sept 2026 for AI Pro and Ultra users only. Here is what that means for visibility and how to test it well.
- Meta Muse Agent on WhatsApp: What Brands Should Prepare18 Sept 2026Meta Muse launched 8 Sept 2026 as a WhatsApp AI agent that can shop for users. What brands, D2C sellers and WhatsApp Business teams should prepare now.
- On-Device AI Agents: What They Mean for Mobile Marketing18 Sept 2026On-device AI agents are reaching flagship phones. What they mean for push, app discovery, search, privacy and mobile marketing, plus a readiness checklist.
- Gemini 3.8 Flash and the Flagship That Never Came12 Sept 2026Gemini 3.8 Flash is Google's best shipped model as of September 2026, because Gemini 3.5 Pro still has not launched. Why that is only half a problem.
- Gemini Hit a Billion Users, That's a Channel Now12 Sept 2026Gemini reached 1 billion monthly users. Gemini 1 billion users marketing means brand visibility inside Google's assistant now operates at search scale.
- Meta Muse: Your Customer's AI Now Does the Buying12 Sept 2026Meta Muse AI agent launched 8 September 2026 on WhatsApp, web and mobile, and it buys. What agentic purchasing means for your marketing funnel.
Choosing, routing and cost 13
Method posts that outlive any single release.
- Governing a Team Prompt Library Without Killing Speed14 Oct 2026A light governance model for a marketing team's prompt library: owners, tiers, versioning, review and retirement, with templates that keep people shipping fast.
- AI Slop Backlash: Will Platforms Demote Generic AI Content?28 Sept 2026Will platforms demote AI slop? What Google's scaled content policy and YouTube's inauthentic content rules say, what is reported, and how to stay safe.
- Which AI Model for Marketing? A Release-Fatigue Framework27 Sept 2026Sonnet 5.5, GPT-6.1 Sol, Gemini 3.8 Flash: new models land weekly. Here is a simple framework for marketing teams to pick, test and stop chasing every release.
- Building an AI Budget Your Founder Will Approve in 202612 Sept 2026Build an AI budget for your marketing team that a founder will sign: cost per deliverable framing, a one-page business case and runaway-spend guardrails.
- AI Model Pricing Compared: The Marketer's Version 202612 Sept 2026An AI model pricing comparison for marketers: input, output and cache rates side by side, and why identical sticker prices hide very different real bills.
- The Model Routing Strategy for Marketing Teams12 Sept 2026The best AI model for marketing 2026 isn't one model. A task-to-tier routing table, implementation steps, and what it saves at real content volume.
- Best AI Model for Content Writing in 2026: A Cost Guide12 Sept 2026The best AI model for high-volume content writing isn't the smartest one. A cost-per-job breakdown of frontier versus Flash-class models.
- How to Evaluate AI Models for Marketing: A Test Protocol12 Sept 2026How to evaluate AI models for marketing with your own briefs, not benchmarks: a blind-test protocol, a scoring rubric and why editing time is the real metric.
- Prompt Caching for Marketing Teams: Cut AI Costs Fast12 Sept 2026Prompt caching for marketing teams in plain English: restructure prompts so brand context caches, cut token costs and stop paying twice for your brand bible.
- The September 2026 AI Model Landscape, Mapped12 Sept 2026Every real AI model released by September 2026, OpenAI, Anthropic, Google, Meta, xAI, Qwen, Mistral, plus the ones that don't exist. Verified, dated, honest.
- What a 1,500-Word Blog Post Actually Costs in 202612 Sept 2026The real cost of AI-generated content, modelled line by line across frontier, Flash-class and self-hosted models, including the editing time nobody counts.
- When Cheap AI Models for Marketing Beat the Frontier Ones12 Sept 2026Cheap AI models for marketing match frontier models on most real tasks. A task-by-task table, falling Flash pricing, self-hosting, and the honest counter-case.
- Why Your AI Bill Went Up When Prices Went Down12 Sept 2026AI costs rising for marketing teams despite falling per-token prices? Where the spend goes, how to diagnose it, and the routing discipline that fixes it.
Agents and automation 22
What agents can safely own in a marketing team.
- Internal Linking Audits With AI Agents: A Safe Workflow14 Oct 2026A safe, human-gated workflow for using AI agents to audit internal links: read-only discovery, proposals, review, staged edits and verification.
- AI Agent Security for Marketing Teams: AgentCore Lessons9 Oct 2026Zenity's AgentCorruption research (8 Oct 2026) reported that one prompt could expose AWS AgentCore agents. What marketing teams should check on their agents.
- What Is an AI Agent? A Plain Definition for Marketers 20263 Oct 2026An AI agent is software where an AI model directs its own steps and tools to reach a goal. Definition, agents vs chatbots, Dots, Agents API. As of 4 Oct 2026.
- An Honest 2026 AI-Marketing Stack Audit: Keep, Cut, Watch29 Sept 2026How to audit your AI marketing stack in 2026: a keep, cut, watch framework with a scoring sheet, cost traps to check and what I'd do in each category.
- Agency Reporting Automation With AI Agents: A 2026 Guide24 Sept 2026How to automate agency client reporting with AI agents in 2026: data pipeline, agent roles, review gates, and the tasks you should never hand to an agent.
- Year End Content Audit: A Practical Q4 Checklist for 202623 Sept 2026A step-by-step year end content audit checklist for Q4 2026: inventory, performance, AI citations, decay, consolidation and a 2027 plan you can act on.
- Agent-Readable Product Data: A Practical Spec for 202612 Sept 2026Structured product data for AI agents, field by field: schema.org Product and Offer, price, stock, shipping, returns, variants, and why it must be in the HTML.
- Agentic Commerce Is Here, But Not How You Were Told12 Sept 2026An evidence-led look at agentic commerce in 2026: what Meta Muse can really buy, why OpenAI retreated once, and the \"discover in AI, buy on site\" consensus.
- Marketing Tasks Worth Handing to an Agent (and Which Aren't)12 Sept 2026A scoping guide to AI agents for marketing tasks: what to automate, what to keep human, the checkpoints you need and the token cost multiplier to budget.
- Building a Multi-Agent Content Pipeline With the Agents API12 Sept 2026A practical multi-agent marketing workflow built on OpenAI's Agents API: real architecture, model tiers, failure modes, and honest cost maths for content teams.
- What Breaks When an AI Agent Visits Your Website12 Sept 2026A practical AI agent website audit: how to fetch your own pages as an agent does, what fails, and the priority fixes worth making before agent traffic grows.
- Agentic Commerce Reality Check: What Actually Works in 202628 Aug 2026OpenAI wound down Instant Checkout. Here's an honest agentic commerce reality check: the protocols, the conversion data, and what to fix now.
- What AI-Agent Layoffs Mean for Your Marketing Career26 Aug 2026Honest India-context guide to marketing jobs AI 2026: which tasks agents actually automate, which skills get more valuable, and a 90-day career plan.
- My Real AI Content Workflow as a Solo Marketer16 Aug 2026A first-person AI content workflow for solo marketers, research, drafting, editing for voice, and repurposing, plus where AI shouldn't decide.
- AI Shopping Agents Checkout Optimization Guide 202613 Aug 2026ChatGPT Instant Checkout and Perplexity's PayPal integration let AI agents buy for users. Here's how to optimize product pages for AI shopping agents.
- Agentic Commerce for Edtech: AI Course Discovery Guide11 Aug 2026Prospective students are researching courses via AI chat before visiting a site. Here's how to structure edtech program data for AI course discovery.
- AI Agents for SEO in 2026: What Works, What Doesn't16 Jul 2026A practitioner's honest review of AI agents for SEO in 2026, what they genuinely do well, where they still fail, and how to use them right.
- Agentic AI Marketing Stats 2026: What the Data Shows18 Jun 2026Agentic AI marketing statistics 2026, real adoption trends, use cases, and ROI patterns, with an honest look at what's hype vs. what actually works.
- Agentic SEO Limitations 2026: What AI Agents Get Wrong16 Jun 2026Agentic SEO limitations in 2026, an honest look at what AI agents do well and where they still fail, hallucinate, or need heavy human oversight.
- AI Agents Replacing Marketing Tools 2026: A Decision Guide14 Jun 2026Should AI agents replace your marketing tool stack in 2026? A practical framework for solo marketers weighing cost, reliability, and control.
- Agentic AI for Admissions Counseling: What to Automate12 Jun 2026Agentic AI admissions counseling explained, how edtech brands automate the first conversation, and where a human counselor still has to take over.
- AI Agents for Lead Scoring: Admissions Qualification 202610 Jun 2026How AI agents score and prioritize admissions leads in 2026, a practical framework for autonomous inquiry qualification without over-automating trust.
Ads automation 4
AI features inside Google Ads and their trade-offs.
- AI Max Search Terms and Ad Combinations Report: A Guide11 Oct 2026Google Ads added a Search terms and ad combinations view for AI Max. Learn what each column means, how to spot mismatches, and a weekly review routine.
- Demand Gen Auto Text Overlays: Controls and Brand Risks11 Oct 2026Google Ads reportedly added an Overlay on videos setting to Demand Gen, possibly on by default. How it works, the brand risks, and a 20-minute audit checklist.
- Google Ads Natural-Language Campaign Setup: Adoption Guide11 Oct 2026Google Ads is testing campaign creation from a plain-English goal. What is confirmed, what is not, and a cautious checklist for trying it with control.
- Google Ads AI Max, Conversion Lift, Target ROAS From Margins8 Oct 2026Google Ads opened Conversion Lift to Search and PMax, is testing a margin-based Target ROAS calculator, and added AI Max previews. What changed and what to do.
More in this hub 1
Everything else filed here.
Also relevant 2
Filed under a neighbouring hub, but worth reading here.
- Best AEO Tools 2026: What I Use With Client Accounts29 May 2026A practitioner's guide to the best AEO tools in 2026, AI-citation trackers, schema generators, and content structuring tools I actually run.
- Best AI Marketing Tools for Solo Agencies 202627 May 2026The best AI marketing tools for solo agencies and small consultancies in 2026, a lean, budget-conscious stack by category, no fluff.
Questions people ask
- Which AI model should a marketing team use?
- There is no single answer. Run the same briefs through two candidates, measure human edit time and cost per finished piece, and route accordingly. The framework and test protocol in the reading path describe how.
- Do model prices in these posts stay current?
- No. Each post states the date its numbers were taken. Treat the method as durable and the figures as a snapshot, and check the vendor pricing page before budgeting.
- Are agents ready to run marketing unattended?
- For narrow, checkable tasks, partly. For anything public-facing or brand-sensitive, I keep a human review step. The posts on tasks worth automating and on agentic limitations explain where I draw the line.
Related hubs
If you want help designing a routing setup around your actual content operation, say what you produce each week.