AND Kimi also moves to the Frontier Tier with the release of K2 thinking
Nathan Lambert
 @natolambert
Aug 17
A tier list of China's top 19 open model builders. Who did we miss? At the frontier * DeepSeek * Qwen Close competitors * Moonshot AI (Kimi) * Zhipu / Z AI Noteworthy * StepFun * Tencent (Hunyuan) * RedNote (Xiaohongshu) * MiniMax * OpenGVLab / InternLM * Skywork On the rise * ByteDance Seed * OpenBMB * Xiaomi (MiMo) * Baidu (ERNIE) Honorable Mentions * Multimodal Art Projection * Alibaba International Digital Commerce Group * Beijing Academy of Artificial Intelligence (BAAI) * inclusionAI * Pangu (Huawei) I learned a lot from these. We have so much more we need to do to understand how their AI ecosystem works.

Nov 6, 2025 · 4:06 PM UTC

18
58
4
940
Replying to @zephyr_z9
i would prefer ZAI models also frontier since everyone is using it to finetune for their own usecases
3
Replying to @zephyr_z9
Team USA is really feeling the pressure now
1
Replying to @zephyr_z9
Might even displace Qwen.
1
36
Replying to @zephyr_z9
I prefer Kimi to qwen. Much better output especially in creative writing. Qwen seems to be too overtrained, every output is 10k tokens for no reason
15
Replying to @zephyr_z9
ByteDance needs to step up their game.
4
Replying to @zephyr_z9
slowly and then all at once 👀
3
Replying to @zephyr_z9
Minimax is also a close competitor in my opinion. Minimax-M2 is very good.
9
Replying to @zephyr_z9
KIMI, GLM, DeepSeek same league. Qwen still cooking r1 style math reasoner and overfitting AIME. Qwen needs to catch up fast!
1
23
Replying to @zephyr_z9
Glm more relevant that deepseek
Replying to @zephyr_z9
Moonshot goes to frontier, MiniMax goes to close competitors
Replying to @zephyr_z9
you're missing a tier for 01[dot]ai
Replying to @zephyr_z9 @natolambert
How is Deepseek still top
Replying to @zephyr_z9
i think so,i use kimi much more after kimi k2
Replying to @zephyr_z9
Maybe you can also pay attention to the Spark 1.5 model
Replying to @zephyr_z9