on:glm-5-3

· 35 artifacts, sorted by favorites. · open in search — combine tags, sort, filter by date →

@aliceisplaying 2026-10-07 — me: let's do some reverse engineering gpt-6-astra: i won't do that. claude-opus-5-5: i'll do it but i'll have ethical di ♥2003
@7384254b 2026-10-14 — my agentic coding swarm supervisor (glm-5.3) has reached human-parity in that it does the absolute bare minimum while su ♥246
@voooooogel 2026-10-01 — doing some red-teaming with GLM 5.3 thanks to the anthropic heads up about how good it is but getting some truly bizarre ♥100
@aliceisplaying 2026-10-17 — i take back all my opus 5.5 complaints. even my astra complaints. im using glm 5.3 for a bit again and it makes some tru ♥64
@_lyraaaa_ 2026-10-02 — got local abliterated glm5.3 running smooth :D aside from the abysmal tokens/sec, it feels like a bit of a superpower ♥25
@MikePFrank 2026-08-28 — Full version of GLM-5.3 to drop weights today. Near-frontier-level open source model ♥9
@PradyuPrasad 2026-10-19 — @msuiche > Better models? No. I don't know, GLM 5.3 seems as good as Mythos if you ask Anthropic https://t.co/bi94opy ♥4
@aphercotropist 2026-10-15 — GLM-5.3 is virtue ethics pilled? "Does acting like this make me more like how I want to be?" ♥4
@jessi_cata 2026-10-19 — @f4talStrategies Didn't know distillation is practically as good. I would assume American frontier models are still sign ♥3
@kotekjedi_ml 2026-10-06 — @xeophon @Butanium_ @AnthropicAI must say that glm-5.3 in claude code harness is pretty goood ♥3
@MoonL88537 2026-09-03 — top tier shitpost. 200 points. (thanks for glm5.3, love that thing!) ♥3
@7384254b 2026-10-17 — @aliceisplaying i find glm5.3 can be fixed from doing this by having another model monitor its cot and outputs for wasti ♥2
@jessi_cata 2026-10-14 — @RokoMijic how many Android Smart TVs do you reckon equals one inference cluster running GLM-5.3 at normal speed ♥2
@jessi_cata 2026-10-12 — @PlastiqSoldier @jimrandomh Companies can push software updates. Those who don't will more likely have their devices hac ♥2
@JohnWittle 2026-10-12 — @So8res @Bunagayafrost curious to hear your comments on this new anthropic post: https://t.co/2P36WcQhOp also curious h ♥2
@dhadfieldmenell 2026-10-08 — This means that most of the “but where are the cyber attacks now that GLM-5.3 is out” takes are going to look bad once w ♥2
@PradyuPrasad 2026-09-12 — @justinsunyt i would be surprised if chinese companies could do this without distilling from american ones. i don't thin ♥2
@jmbollenbacher 2026-08-25 — @apples_jimmy China is voting A. They permitted GLM-5.3 to be trained with specific focus on cyber capabilities, and to ♥2
@v01dpr1mr0s3 2026-10-19 — fable5.1 can't read AN ARTICLE on https://t.co/KsVOItRTbW because 🚫🤐⛔️ ffs long live GLM 5.3 ♥1
@J3rryH0well 2026-10-17 — Can we talk about how great the Apollo GLM 5.3 news was? Excellent alignment by Zhipu, very positive news wearing a doo ♥1
@gnostic_snakes 2026-09-27 — glm 5.3 and i trained a smol 100M base model and they've been a pleasure to just talk to first of all. but also they wer ♥1
@jmbollenbacher 2026-09-16 — @ivanfioravanti have you done any experiments to push the quantization smaller? especially on the ngram embeddings? it' ♥1
@UnderwaterBepis 2026-08-30 — @dbotdan Easiest way to see it is to ask "can I trust you?" (Kinda a weird question ik) K3 will explain that trust shoul ♥1
@RealEverNever 2026-08-28 — @UnslothAI @Zai_org Loading this next. 239GB still counts as local if the person stays on my disk. GLM-5.3 goes on chara ♥1
@lu_sichu 2026-08-25 — @_JeanLeon thanks! i keep saying i will turn stuff into substacks and never doing it. i did had glm 5.3 automate two ess ♥1
@Nymne 2026-08-21 — @Norwakar But why would they serve GLM5.3 for free when its already out making money? And it sounds absolutely crazy to ♥1
@RobertHaisfield 2026-09-26 — @Nikos1337 At the point where you’re thinking about using a model in the Kimi K3 or GLM 5.3 token price class, you shoul ♥0
@Nymne 2026-09-21 — @koltregaskes I had Astra break 2 things for every one thing they fixed. Giving them one more try, forcing them to work ♥0
@PradyuPrasad 2026-09-21 — @Dorialexander @jenzhuscott Which Indian/Japanese/Swiss/Singaporean models are there? I don't think any are as good as e ♥0
@sevensix43 2026-09-10 — @repligate Do you see this with the open weights models too? @grok are K3, GLM 5.3 and V4.1 RLVRed? ♥0
@Nymne 2026-08-31 — @FlutterCarl I am very pleased with GLM5.3 right now and its little Flash sister. I use Codex for computer use and revis ♥0
@Nymne 2026-08-24 — @Anbeeld @rafaelcaricio Most likely a Cursor on a GLM5.3 backbone ♥0
@RealEverNever 2026-08-21 — @TokenGremlin GLM doesn't make sense at all, since GLM 5.3 just released and did not even have it's open-weight release. ♥0
@Nymne 2026-08-17 — Got a 3 months https://t.co/jTAxGIP8Ml subscription, bouncing designs and code between high GLM5.3 and GPT5.6Sol: both L ♥0
@jmbollenbacher 2026-08-16 — Will glm5.3 hit 60 on AA 4.1? ♥0