Gergely Orosz argues surging GLM-5.2 adoption threatens OpenAI and Anthropic revenues as open-source matches closed-source coding performance

This is from a popular inference provider

GLM-5.2 plus the US banning the most capable new models means open source caught up to SOTA closed source coding models

This could be v problematic for Anthropic’s and OpenAI’s revenue projections and IPO plans…

@Madisonkanna open source is more important than even in this moment. do you all publish benchmarks to show how glm is performing for you? i'd be happy to amplify if you do.

@MatthewBerman that's so kind of you, we should collab sometime. also yes! check out @philipkiely's article :

@Madisonkanna it stands for Good Luck Madison

@paularambles Oh my gosh I laughed out loud

@Madisonkanna absolutely tectonic shift, exciting time to be building on top of open models

@Madisonkanna not me who keeps reading GLM as GLP 😭😂

@Madisonkanna lets go

@Madisonkanna Pls

@MatthewBerman @Madisonkanna @MatthewBerman at this point benchmarks are self-reporting. If they pulled Fable and GPT-5.6 for being powerful models and an open source model comes out reporting that it’s on par or even better than those models, no way the Gov will let it remain public.

@thealexker I’m excited

@Madisonkanna @philipkiely 👀👀

@Madisonkanna love it

@Madisonkanna the biggest winner isn't one model.

it's builders getting more real choices.

@Madisonkanna as someone who has made a trillion+ parameter model go 300+ tps, what do you make of the 750 tps claim for OpenAI Sol?

@Madisonkanna 🎉

@Madisonkanna The requests will increase, we've no choice.

@Madisonkanna This is great! US is crafting a footgun for themeselves and national AI companies

@GergelyOrosz As we say in Kentucky: Bless their hearts.

They sowed these seeds years ago with the alarmism marketing strategy

@Madisonkanna USG is screwing the American people.