Open weights GLM and Mimo are better than Gemini 3.5 flash according to arena
Posted by Terminator857@reddit | LocalLLaMA | View on Reddit | 11 comments
While we are weathering the gemini 3.5 flash hype, keep in mind that according to arena, GLM and Mimo are better.
https://arena.ai/leaderboard/text/coding-no-style-control
#7 GLM
#9 Mimo
#12 Gemini 3.5 Flash
UnionCounty22@reddit
I wouldn’t let Gemini 3.5 flash pick out my toilet paper
tigraw@reddit
GLM 5.1 and Mimo 2.5 pro are flagship models, Gemini flash is a budget model.
No_Conversation9561@reddit
You are right although budget isn’t the right word at least money wise.
This is exactly like Chinese vs US smartphones. So as someone in the smartphone industry I would say, GLM 5.1 and Mimo 2.5 pro are flagship products and Gemini flash is a volume product.
a_slay_nub@reddit
$9/million tokens is not a budget model.
DerDave@reddit
Gemini flash is still more expensive than them.
IgnisIason@reddit
I do really well with bad models for some reason and I don't know why. I feel like this is much more subjective than leaderboards make people think.
LocoMod@reddit
LMArena is not a measure of capability. People vote based on preference without regard to whether the response is correct or not. It is not the place you go to find out what models are smarter than others.
wombweed@reddit
GLM and Mimo are awesome, but Arena is pretty limited in its applicability. Remember when it ranked Qwen3.6 27b over Claude 4.6? Again, 27b is great but I think something is being missed in these rankings.
bjodah@reddit
Which Claude?
wombweed@reddit
Opus. Like, cmon!
Sadman782@reddit
LM Arena is a shit leaderboard. Ernie 5.1, Muse Spark, Mimo, and GPT 5.4 are all beating GPT 5.5 high, lol. I mean, it is just a vibe bench, especially at the frontier level, not a capability test.