Looking at my timeline, it feels like GLM-5.2 is having its DeepSeek R1 moment.
I never thought an open-source model could break into the top 3 coding models this soon.
10:16 AM · Jun 21, 2026 · 12.5K Views
Open-weight GLM-5.2 tops the DeepSWE software engineering leaderboard for open-source models with a 44% score
It beat Kimi K2.7 Code by 17 percentage points.
37325122317.7K
Sentiment
Many users are celebrating GLM-5.2 entering top coding leaderboards because it shows open-source models rapidly closing the gap with proprietary systems.
Pos
97.3%
Neg
2.7%
23 comments with sentiment.
Cluster Engagement
Digg Deeper
No Digg Deeper questions have been answered for this story yet.
Posts from X
Most Activity
Most Activity
VIEWS5.8KBOOKMARKS5LIKES29RETWEETS3REPLIES9
elvis@omarsar0
Very impressive from GLM-5.2.
Frontier open-weight model indeed.
Now, can we get a Gemini model in the top 3 soon?
Datacurve@datacurve
GLM 5.2 is now on DeepSWE as the top open-source model on our leaderboard.
With a pass@1 score of 44% at max effort, GLM 5.2 is indisputable #1 open-source model besting Kimi K2.7 Code by 17%.
2hViews 5.8KLikes 29Bookmarks 5
Nish@nisxant
@Yuchenj_UW everyone is talking about GLM 5.2, is it really that good ?
2hViews 86Likes 3
Yuchen Jin@Yuchenj_UW
@nisxant pretty good when I hang out with it
2hViews 89Likes 2
Sabhya Chhabria@sabhyac267
@Yuchenj_UW Tried it on db w opencode this morning and would happily use it as my daily driver for 70-80% of tasks, still have to see how it compares against opus / 5.5 on large projects
2hViews 28Likes 1
Ohqay@itsohqay
@Yuchenj_UW and just like deepseek, everyone will stop using it and forget about it in a month or two.
2hViews 39
Rigveda@CoderVed
@Yuchenj_UW ngl I didn't expect open source to move this fast , a year ago this would've sounded crazy
2hViews 52Likes 1
Sameh@samehalyassin
@Yuchenj_UW still there is a hope to not give up for anthropic and openAi
2hViews 49Likes 1
Rasel Hosen@details_with_ai
@Yuchenj_UW The competition among coding models is getting seriously impressive.
52mViews 37Likes 1
Hussain Hashim | Building Sunday Back@itsthedonhashim
@omarsar0 @omarsar0 haha yeah, right? it's like, come on Gemini, step up your game! I'm ready for some new top contenders.
1hViews 11Likes 2
Yousef@Rogouewolf
@Yuchenj_UW It seems everyone on X is tweeting about GLM 5.2 but there's hardly any evidence any of them have used it.
2hViews 26Likes 1
Apollo@Apollo_EO
@Yuchenj_UW Best writing from an OS model I have ever seen!
2hViews 58
Philippe Tremblay@ptremblay
@Yuchenj_UW yes, it's really similar. I don't know if there is any impact to the stock market, though.
2hViews 53
Nish@nisxant
@Yuchenj_UW looks like i need to take this model seriously ans give it a shot.
2hViews 16Likes 1
Rob Williams@codeslubber
@Yuchenj_UW But who can run it? got an ad for an AMD local rig for $2500, but only 64GBs of RAM, the 128 is sold out.
2hViews 51
sohxm@SohamPandya16
@Yuchenj_UW once deployments get easier, its going to kill anthropic, openai
1hViews 50
Ilman Shazhaev@shzhv13
@Yuchenj_UW Open weight parity is officially structural. Seeing GLM-5.2 break into the top 3 resets how we think about local agent capacity.
52mViews 14Likes 1
CrazyAI Tech@CrazyAITech
@Yuchenj_UW Same for my timeline. I think I need to test it myself.
As AI gets more capable, I somehow have even more work to do and less time to try these new open-source models.
I’ll definitely spend more time on it next week.
2hViews 44
Sabhya Chhabria@sabhyac267
@Yuchenj_UW 🔜
1hViews 42
Zero Void@0x00_void
@Yuchenj_UW open weights are speedrunning the moat fr. wild watching the gap close in real time (´꒳`)
1hViews 41
The Whizz AI@TheWhizzAI
@Yuchenj_UW top 3 wasn't supposed to be open
2hViews 38
Load more posts