Almost Timely News: 🗞️ 4 Angles on Local AI (2026-06-14)
Almost Timely News: 🗞️ 4 Angles on Local AI (2026-06-14)A fable is a short fictitious story designed to teach a specific moral lesson.
Almost Timely News: 🗞️ 4 Angles on Local AI (2026-06-14) :: View in Browser
The Big Plug👉 My new course, GEO 201 on competitive GEO measurement, is now for sale. Content Authenticity Statement100% of this week’s newsletter was made by me, the human, and boy does it show. Learn why this kind of disclosure is a good idea and might be required for anyone doing business in any capacity with the EU in the near future. Watch This Newsletter On YouTube 📺
Click here for the video 📺 version of this newsletter on YouTube » Click here for an MP3 audio 🎧 only version » What’s On My Mind: 4 Angles on Local AIThis week’s been one of those weeks where so much has happened and it’s so messy that there isn’t a theme, so I’m going to foam at the keyboard, write down my thoughts, and see what comes out on the other side. Part 1: A Fable of Fable“A fable is a short fictitious story designed to teach a specific moral lesson.” This week, Anthropic’s fifth generation model family, Fable, became available for a short period of time. It debuted on Tuesday and was blocked by the US government on Friday for ambiguous reasons without clear evidence, despite the fact that its larger, more dangerous version, Mythos, has been available to large corporations for a few weeks now. In my tests of Fable while it was still available, it was excellent at what it did, very, very expensive, and has clear use cases. What I found interesting was the level of amazement people had at it - folks were raving over its abilities to discern intent from ambiguous, poorly written prompts and turn them into real results. While that’s admirable of Fable’s capabilities, it speaks more towards people’s unwillingness to take the time and plan and think things through. Opus 4.8 on xHigh or Ultra settings can accomplish most of what Fable can do - if you prompt it well and let it iterate and think and review. Here’s my general process for this, something Katie Robbert taught me. First, never just go off and do something. Take the time to build out the requirements, ideally with a framework like the 5P Framework by Trust Insights™. Once you’ve got requirements nailed down, build a specification. This can be a design spec, a writing spec, a code spec, something that says “here’s what we’re doing”. Then build a workplan from your spec. Finally, and only after you have requirements, spec, and plan, do you have AI go off and do the thing. If you follow this general recipe, not only do you get great results, you also don’t have to use the biggest, heaviest, most costly AI model to do it with. By the time you’ve reached the workplan and reconciled it with the requirements and spec, you’ll have anticipated the majority of things that could go wrong. Where I found Fable quite powerful was in review; I had it delegate lighter tasks to smaller sub-models but have it be a master review agent at the beginning and ending of a process, uncovering lots of bugs and issues that previous QA runs hadn’t picked up or seen as a coherent picture. In that regard, it impressed the heck out of me. But the fable here is something I wrote about earlier this year - if you want guaranteed access to AI, you absolutely must have a private, local version running somehow. It can be on your own hardware if you have a bespoke machine like an Asus GX10 or NVIDIA DGX Spark, or a well appointed Mac. It can be on your company’s hardware. It can be on a bespoke hosting service, ideally outside your local jurisdiction, but no matter what solution you pursue, you had better have a fallback. The events of this week proved that any government can unilaterally cut you off from AI services and because cloud providers must adhere to lawful requests, your access to cloud AI is contingent on your government’s approval. Having your own as backup isn’t a nice to have, not if your business relies on generative AI now. The AI policy nerds call this AI sovereignty; every country should have its own AI so that no one country or government has control over it. More on this in a bit. Part 2: Mini-token-maxxingA couple of weeks ago, Minimax M3 came out and the company changed its billing for their token plan. Previously, for the M2.7 model, they billed by request, which was ideal for AI agents like OpenClaw and Hermes Agent. When M3 came out, they switched it to token billing, allotting 1.7 billion tokens per month on their Plus plan. That sounds like a lot, but it’s really not. How much not? When I was doing some work with my agent earlier this week, I kept hitting the 5 hour usage limit wall. Curious, I switched its operation from Minimax to Qwen 3.6 running on my MacBook (yay local AI). What I found shocked me - Hermes Agent was churning through about 13 million tokens an hour. Now, most of those were cached, meaning that the processing load was relatively small, but Minimax bills by token whether it’s cached or not. If you haven’t worked with pay-as-you-go AI services (like Claude Enterprise, for example, or the many API versions of common AI tools), there’s usually a difference between new tokens and cached tokens, in terms of pricing. Cached tokens usually cost significantly less than new ones, because if you can re-use cached tokens, you cut your costs. Here’s a simple example. Imagine you commission a ghostwriter to write a blog post for you. The ghostwriter bills by the hour. After the first paragraph, you have a chat because you want the post to go in a different direction. If the ghostwriter has to start from scratch, you’re going to pay full price for their output. If, on the other hand, the ghostwriter can reuse the paragraph they’ve already written - a cached version, then you don’t have to pay for that time and output again. That’s what cached tokens means - AI already has some output it can reuse. Normally, in API versions of AI, this is a discount, but under Minimax’s new plan, cached tokens are the same as regular ones. The ghostwriter charges you again for a paragraph they’ve already written and you’ve agreed it can reuse. So my little agent churning away at 13 million tokens an hour - when I plugged it into Qwen, I saw that almost 90% of the tokens were cached. If I was on a regular pay-as-you-go service, I would be spending very little. But instead, on the current token plan from Minimax, I’m using up my quota whether or not the tokens are new. Why are they doing this? Because token plans - and by these I mean any flat rate plan, like Claude Max, ChatGPT, etc. - are money losers for AI companies, often substantial ones. If you look at Claude Max 20 and how much output you get for $200, you’re getting roughly $8,000 worth of tokens. That’s a 97.5% discount - and there’s no way that’s sustainable in the long term. Minimax likely ran into token economics faster than Anthropic, so the new plan with the new model is more realistic in terms of what you pay versus what you get. Expect this to be the case across the industry in the coming months and years. No business can reliably sell its product at a 97.5% discount for long. Minimax’s token plan used to be the best deal in AI; today, it’s a good deal but not a great one - and what you can run on your machine can match it. Also in the news this week, Minimax M3 became open weights; anyone with enough hardware can download it and run it for free, but you need lots and lots of hardware. It’s a beefy model, 428 billion parameters with 23 billion active parameters; a little smaller than DeepSeek, a little larger than Qwen 3.6. M3 is a highly capable model on par with Opus 4.6 for most tasks. A colleague asked this week why these companies are giving away their models. The answer is that China, which heavily subsidizes its AI companies as well, doesn’t want the US to have a monopoly on AI - and with the US government blocking Fable 5, that concern is very real. By giving away their models, they ensure that their technology is broadly available and has no linchpin, no single shatter point. That’s the heart of AI sovereignty, and it’s what China in particular is providing to the rest of the world by giving away its best in class models. And these models aren’t just for the hardware-rich (i.e. folks who can afford to drop tens of thousands of dollars on heavy duty compute) - a model like Qwen 3.6 35B-3AB runs on a well appointed MacBook Pro. According to OpenRouter’s statistics (which measure mostly heavy developer usage of AI, not consumer usage), more than 40% of the planet’s AI traffic uses Chinese models of some kind because of their substantially lower costs. Part 3: Fat DocumentsA few weeks ago, Andy Crestodina and I did a piece together on document formats and AI, and how much more resources some documents use compared to others. At the time, I said my piece as well published my own longer version on ways to save token budgets, but this week I decided to quantify it. I plugged Claude Code into my local AI server and gave it a Word document and the equivalent Markdown format to process. In both cases, I gave it a super naive prompt - “summarize this document”. The backend model for both was Qwen3.6-35B-3AB, running on my MacBook. My local AI summarized the document in 28,693 tokens, of which 20,480 were cached, at 71.4% efficiency. That’s fast and lightweight. The exact same document in Word DOCX format? 52,163 tokens, of which 26,624 were cached, at 51% efficiency. The Word document was 81% heavier, and 28% slower and less efficient. That’s for a single document on a local model; now imagine hundreds or thousands of employees dragging and dropping Word documents, Powerpoints, and Excel spreadsheets into AI all day, every day. You’re talking an unnecessary 81% increase in token usage. And if your enterprise pays by the token, those savings could be considerable. But I wouldn’t have known how big the savings were until I used my local setup to watch in real time as I dropped documents in for processing. That’s another advantage of local AI - you can see exactly what’s happening, blow by blow, in your AI system of choice. Imagine going to your AI Council and telling them, at the enterprise level at least, that you could reduce token usage by 45%. That’s not pocket change, not if you’re paying by token. You could do that today by helping team members convert documents to open, text-based formats like this:
My test this week shows that this works and is highly effective for saving token budget. I would never have seen that if I didn’t have control of the model itself. Part 4: Data Centers and Responsible AIDuring an event I was speaking at this week, an audience member asked for my opinion on data centers and responsible AI. My response was the one I always give: First, use AI as little as possible. Use it for what it’s best at, but know that it LOVES to reinvent the wheel, and it’s in AI providers’ economic incentives for it to do this. The way I think about AI usage today is to use it to build tools (coding is one of its most capable use cases) that then run without AI. Here’s a simple example. Suppose you were using AI with some data from your Google Analytics. If you handed some of your data to AI and asked for an analysis, it would do a very capable job, write a bunch of Python code behind the scenes, and then give you the answer. Very likely, the answer would be correct. Then the next month, when it’s time to write that report again, you ask AI for help and it repeats the same process - but it doesn’t fish out the code it wrote behind the scenes to process the data last month. No, it writes it all over again, completely reinventing the wheel. That’s not how I use AI. Instead I think through whether the outcome is something that generative AI can do or whether it’s something regular code can do - and if it’s the latter, then I’ll spend time having AI build the code so I can reuse it month after month and never have to use AI for the task. Even for something where AI is the right choice, I try to build as much non-AI infrastructure as possible. For example, I start my days by asking Claude inside Claude Code to give me my daily briefing. Under the hood are 4 different command line applications that talk to systems like Asana, GMail, Jira, and others - and none of those applications use AI at all. AI parses the results, but it doesn’t go and fetch data. It doesn’t need to, and it’s not well suited for that. And I’m starting to test swapping out my Qwen model for this particular task because it’s just summarization. Now that the tools all exist, I don’t need to use someone else’s cloud AI for what’s essentially just summarization and prioritization. The version of Qwen I use, per Artificial Analysis, is on par with the agent skills of Sonnet and exceeds Haiku:
Second, if you’re concerned about the power and water consumption of data centers, use local AI as much as you can. I’ve switched over more and more of my AI usage to my MacBook running Qwen 3.6 locally. Why? Because the electricity that powers my laptop comes from the roof of my house (solar power). There’s no fresh water usage at all; if there was, it’d be a disaster because it would mean my laptop was wet and my office was flooded. Data centers have real impacts on the communities they’re located in, drawing power and water - and different companies have different approaches to sustainability. If you care about that, read the various providers’ annual reports on sustainability and choose a provider based on which company is doing the most to lessen its impact. If you’re in a position to do so, run your own AI as much as you can, for as many tasks as you can. Every API call you make to your own hardware is power and water you’re not consuming from somewhere else; even if your own electricity isn’t green, you’re drawing it from your local grid and not having a data center draw it from a very concentrated location. But even if you don’t care about sustainability, conservation here means literally saving money, especially in the enterprise. Every token that’s generated by a cloud AI provider is a token that costs you money. Every token you generate on your own infrastructure is a token you probably paid less for. If you were to look around the average enterprise, I’d wager most of the compute power of that enterprise is sitting idle, most of the time. For example, at Trust Insights, I know most of my coworkers work really hard. They’re smart people. They do all kinds of different tasks. But I’d bet their GPUs and NPUs on their Macs are near idle almost all day long. What if you could harness that power? Well, you can. There are projects like Exo that network together computers on the same physical network (like your office LAN) and soak up that spare computing power to create an AI supercluster. Right now it works best on a LAN, but the community is trying out different options to see about making it work on a WAN for remote workers. Wouldn’t it be cool if you could create an AI supercluster that served most of your AI needs, powered by the computers and the hardware you’ve already paid for? That’s what the future of AI could look like for enterprises that can put together their the hardware they already own. Part 5: Wrapping UpI suppose the throughline for all of today’s hodgepodge of items was about local AI, about its benefits for AI sovereignty, sustainability, business continuity, and more. Local AI is going to become more and more important as we see more issues arise with government control of AI, with sustainability, with token budgets. Having AI that is under your control, on your premises, on your physical hardware, or in your bespoke data center providers will give you those advantages of having reliable AI that cloud providers just can’t promise because of all these external factors. Go back to the archives, and I’ll put some links below for all the issues on how to set up local AI. If your company or you as a professional are reliant on generative AI. You have an obligation to have local AI on tap for business continuity, for sustainability, and for cost reduction. How Was This Issue?Rate this week’s newsletter issue with a single click/tap. Your feedback over time helps me figure out what content to create for you. Here’s The UnsubscribeIt took me a while to find a convenient way to link it up, but here’s how to get to the unsubscribe. If you don’t see anything, here’s the text link to copy and paste: https://almosttimely.substack.com/action/disable_email Share With a Friend or ColleaguePlease share this newsletter with two other people. Send this URL to your friends/colleagues: https://www.christopherspenn.com/newsletter For enrolled subscribers on Substack, there are referral rewards if you refer 100, 200, or 300 other readers. Visit the Leaderboard here. ICYMI: In Case You Missed ItHere’s content from the last week in case things fell through the cracks:
On The TubesHere’s what debuted on my YouTube channel this week: My Merch ShopI’ve been adding so much stuff that I’ve decided to bundle it all in what I call a Merch Shop, because otherwise there’s literally too much to keep track of and I run out of space in my own newsletter. So welcome to the Merch Shop! Skills for Claude and Agentic AI: Books: Courses: Subscriptions: Recent TalksThese are just a few of the classes I have available over at the Trust Insights website that you can take.
Advertisement: New GEO 201 CourseIn GEO 101, the first course I built on the basics of GEO, I taught you about presence, appearance, and relevance, the three phases of GEO, and what you need to do in each phase to align with how AI search operates. The top piece of feedback we got at Trust Insights about it was, “okay, great, but how do I tell my boss that we’re ‘winning’ at GEO?“ After I quelled my murderous rage at your boss on your behalf, Katie and I sat down and worked out a straightforward, aligned methodology for doing this. GEO 201 is based on the three phases, what you can control and what you can genuinely see - and critically, what you can’t. Because there is absolutely no way to say your brand “ranks higher” in AI search, period, end of story. But you can say and show with confidence what you’ve done and how you show up for presence, appearance, and relevance with tools you’re probably already paying for, and based on how AI search systems really work. 👉 GEO 201 is available now for USD 149. Get Back To Work!Folks who post jobs in the free Analytics for Marketers Slack community may have those jobs shared here, too. If you’re looking for work, check out these recent open positions, and check out the Slack group for the comprehensive list.
Disclosure: I source these links from LinkedIn every week on the following criteria: New in the past seven days, Easy Apply on, remote roles, USA geography. How to Stay in TouchLet’s make sure we’re connected in the places it suits you best. Here’s where you can find different content:
Listen to my theme song as a new single: Social Good: Ukraine 🇺🇦 Humanitarian FundThe war to free Ukraine continues. If you’d like to support humanitarian efforts in Ukraine, the Ukrainian government has set up a special portal, United24, to help make contributing easy. The effort to free Ukraine from Russia’s illegal invasion needs your ongoing support. 👉 Donate today to the Ukraine Humanitarian Relief Fund » Events I’ll Be AtHere are the public events where I’m speaking and attending. Say hi if you’re at an event also:
There are also private events that aren’t open to the public. If you’re an event organizer, let me help your event shine. Visit my speaking page for more details. Can’t be at an event? Stop by my private Slack group instead, Analytics for Marketers. Required DisclosuresEvents with links have purchased sponsorships in this newsletter and as a result, I receive direct financial compensation for promoting them. Advertisements in this newsletter have paid to be promoted, and as a result, I receive direct financial compensation for promoting them. My company, Trust Insights, maintains business partnerships with companies including, but not limited to, Amazon, Talkwalker, MarketingProfs, Agorapulse, The Marketing AI Institute, Spin Sucks, and others. While links shared from partners are not explicit endorsements, nor do they directly financially benefit Trust Insights, a commercial relationship exists for which Trust Insights may receive indirect financial benefit, and thus I may receive indirect financial benefit from them as well. Thank YouThanks for subscribing and reading this far. I appreciate it. As always, thank you for your support, your attention, and your kindness. Please share this newsletter with two other people. See you next week, Christopher S. Penn Invite your friends and earn rewards
If you enjoy Almost Timely Newsletter, share it with your friends and earn rewards when they subscribe.
|
Comments