Every Trick to Make Local AI Faster: the Cheat Sheet
My code and writing speeds, plus the Mac and PC settings to copy
For: Anyone using local AI on an Apple silicon Mac or a PC with a graphics card.
I gave five setups the same physics simulation to code and the same story to write. On my MacBook Pro M5 Max with 128 GB, plain GGUF wrote the code at 25.2 tokens per second. Splash wrote it at 97.3.
This free sheet goes with Every Trick to Make Local AI Faster. Page one puts the two tasks side by side. The other pages show which model to download, where to turn on its built-in guesser, how to pair a DFlash draft, and which Mac models Splash supports. On a PC, start by checking what fits in graphics memory. On either platform, match the thinking level to the job.
Unlock this cheat sheet
Enter your email and the unlock link lands in your inbox. One signup opens every free resource on this site, and you get The Signal, our free newsletter on putting AI to work. Unsubscribe anytime.
My measured speeds at a glance The full cheat sheet (PDF, 3 pages)
We use your email to deliver the unlock link and The Signal. Details in the privacy policy.
Check your inbox
The unlock link is on its way to . One click there and everything on this page opens.
Nothing after a couple of minutes? Check spam or promotions.
While you wait: what best describes your work?
Noted. The link should be with you shortly.
Going deeper than free tools? See the Build With AI guide →
Want the next one first?
New tools like this go out in the newsletter, alongside the workflows and tools actually worth your time. 2x a week, no noise.
Free. Unsubscribe anytime.
Ready to go deeper?
Build With AI
Direct AI to build products for you — like having your own software engineer. No coding required.
Explore the guide — $179Want 1:1 help putting AI to work? See coaching options →
More free resources
Ollama vs LM Studio vs oMLX: the Free Cheat Sheet
Where each setting is in all three apps, and every command to copy
Buying a Mac for AI: the 5 Catches Checklist
Five checks before checkout, with measured numbers and a Mac and memory lookup by workload
Before You Buy a Computer for AI: The 5 Checks
What to check on any Mac or PC before you spend thousands on it for AI: the memory your jobs need, how fast is fast enough, the five checks, the ~$3 test, and whether to buy now or wait
4 Ways to Run AI Bigger Than Your Memory
The Mac setup, the command for files you already have, and the way that fits your machine
The Best Local AI for Each Job: the Free Cheat Sheet
Twelve everyday jobs, one free pick for each: the app it runs in, the download, the memory it needs, the settings and what the demo showed
Claude Max vs ChatGPT Pro: If You Can Only Choose One
Every personal plan on one page, with the cutoff that matters at each price
Local AI at Every Memory Size: the Cheat Sheet
From 16 GB to 1 TB: what you can buy, how much the AI really gets, which models fit, how smart they are and how fast they write
The 128 GB Local AI Cheat Sheet
Every way to get 128 GB, what each one costs, how fast it runs, and what to buy at each budget
GPT-5.6 Settings Cheat-Sheet
Use the best model of the year without torching your usage limits
The AI Automation Toolkit
Find what to automate in your life, then build it in plain English
The PM Prototyping Kit
Go from an idea to a clickable, shareable prototype in a day — with Claude Code