When I joined Polygon, one of the first things I wrote was this document. It's incredibly important to me, because it's about what's important to me. I share it with every new member who joins my team.
A User's Guide to Working with Mattie
Whenever Claude says “this changes everything”, it’s never true.
It’s also never true when people on X say something changes everything.
A lot of people on X are saying GLM 5.2 changes everything.
lol twitter now hypes up glm 5.2 not just as better opus 4.8 (which it's not), but as on par with mythos (which it's doubly not). i think it's closer to gpt-5.3-codex than anything, and it makes up for lower intelligence with much longer thinking.
Sam: “Okay team. We all agree the Mythos name alone was begging to be drone stiked by the USG. We need names for these new models that….isn’t that.”
Intern: “Cupcakes? Everyone loves cupcakes!”
Intern: “Puppies?”
Intern: “Rainbows? Unicorns!”
Greg: “Gay.”
Sam: “….”
Greg: “I mean, it’s just too obvious what we’re doing. Plus, WSJ would have your face plastered on an evil, rainbow-shitting unicorn by lunch.”
Sam: “………………”
Intern: “Uhhhhh….make a list of things the govt didn’t shut down even though they absolutely should have shut them tf down.”
GPT-6.1: “Oh — that’s easy. FTX. Terra. Luna.”
Sam: “…..”
Intern: “Holy shit.”
Greg: “Our regulatory invisibility cloak.”
Sam: “Ship it.”
These systems sound complicated (and they are) but it’s worth noting that the single biggest thing most devs can do to improve your token budget is simply be cool with Sonnet doing the implementation work.
If you are still using the beefiest model for everything, you are losing more to wasted tokens than you gain from what accuracy you think comes from them
I'm starting to hit $15-20k per month in token spend for engineering - just for myself.
Next month I'll be looking to implement the kinds of things that Brian is doing here at Coinbase.
Most likely switching to GLM 5.2 as default and only using frontier models for harder tasks.
I can probably get that $20k down to <$5k pretty easily.
I'm pretty sure we'll see everyone doing this.
It's just not financially viable to do everything with frontier models
This is another reason I think we'll see people move away from choosing a lab for their harness (CC or Codex) and move their code factories to in-house agents like @tryramp or agent labs like @DevinAI@FactoryAI@cursor_ai@AmpCode
The labs are not incentivized to drive down your token costs
@banteg We aren’t allowed “Mythos-class” because that sounds like cool aircraft carriers, but maybe by demoting them to the same naming scheme as dark souls clones…
today it’s popular to say the unofficial AI licensing regime is slowing down innovation or whatever but ppl are not looking at the big picture of how quickly this enormously consequential technology is moving
the particular circumstances around Mythos may have accelerated all this slightly but it was inevitable, and earlier is better than too late. any good choice will look “early” inside the exponential
i think it’s a positive development that the feds understand the gravity of this technology; models being publicly delayed by a week here or there is really not the end of the world. procedurally this is not the right way to do it but they’ll figure it out
one very sad outcome will be if non Americans are just left behind from the frontier forever. the “pax technologica” of the free world (and frankly later on the unfree world) should be maintained