Wang Magic Profile
Wang Magic

@WangMagic_

Followers
605
Following
123
Media
3
Statuses
74

RL @theMetaStoneAI, USTC Phd

Joined March 2025
Don't wanna be here? Send us removal request.
@WangMagic_
Wang Magic
21 days
Google’s Gemini2.5 Deep Think claimed they used parallel thinking, and we just publicly released this core technology.
@theMetaStoneAI
MetaStoneAI
21 days
🚀 Introducing XBai o4:a milestone in our 4th-generation open-source technology based on parallel test time scaling!.In its medium mode, XBai o4 now fully outperforms OpenAI−o3−mini.📈. 🔗Open-source weights: .Github link:
Tweet media one
Tweet media two
0
5
36
@WangMagic_
Wang Magic
9 days
RT @AashiDutt: 🚀 Just published: “XBai o4: A Guide With Demo Project”. Walks you through locally deploying @theMetaStoneAI's new reasoning….
0
1
0
@WangMagic_
Wang Magic
16 days
Big move.
@moskstraum21745
moskstraumen
17 days
@WankyuChoi Try and play with Xbai-o4 ("high" setting). While smaller, it outperforms (at the FOMO riddle) a model that's more than 10X its size (GLM-4.5). TTS works.
0
0
6
@WangMagic_
Wang Magic
17 days
Make the community move!.
@sama
Sam Altman
18 days
gpt-oss is a big deal; it is a state-of-the-art open-weights reasoning model, with strong real-world performance comparable to o4-mini, that you can run locally on your own computer (or phone with the smaller size). We believe this is the best and most usable open model in the.
0
0
4
@WangMagic_
Wang Magic
19 days
News: @OpenRouterAI is considering to add XBai o4!
Tweet media one
4
1
29
@WangMagic_
Wang Magic
21 days
Thanks Ivan, anyone can enjoy our XBai o4 in this way.
@ivanfioravanti
Ivan Fioravanti ᯅ
21 days
XBai o4 ready to go on MLX in 4,5,6,8 quantization formats. DWQ cooking. I love the thinking trace of this model. M3 Ultra 80 cores: 30.9 tokens/sec.M4 Max 40 cores: 23.5 tokens/sec. Some context size tests and mmli pro benchmarks in the queue.
Tweet media one
2
1
26
@WangMagic_
Wang Magic
21 days
We recently collected some good cases of XBai o4. The top three most-liked comments on this post will receive 100 million tokens of XBai o4 API credit.
@theMetaStoneAI
MetaStoneAI
21 days
🚀 Introducing XBai o4:a milestone in our 4th-generation open-source technology based on parallel test time scaling!.In its medium mode, XBai o4 now fully outperforms OpenAI−o3−mini.📈. 🔗Open-source weights: .Github link:
Tweet media one
Tweet media two
15
2
53
@WangMagic_
Wang Magic
21 days
XBai o4 has released the core technology of Gemini 2.5 Deep Think in “parallel thinking”🔥@GoogleDeepMind.
@theMetaStoneAI
MetaStoneAI
21 days
🚀 Introducing XBai o4:a milestone in our 4th-generation open-source technology based on parallel test time scaling!.In its medium mode, XBai o4 now fully outperforms OpenAI−o3−mini.📈. 🔗Open-source weights: .Github link:
Tweet media one
Tweet media two
1
1
15
@WangMagic_
Wang Magic
1 month
RT @TheTuringPost: 7 notable AI models to pay attention to:. ▪️ Kimi K2.▪️ Gemini 2.5.▪️ Grok 4.▪️ Voxtral.▪️ MirageLSD.▪️ LTX-Video.▪️ Met….
0
35
0
@WangMagic_
Wang Magic
1 month
human agent based on chatgpt🔍.
@isafulf
Isa Fulford
1 month
tip for chatgpt agent slides: . first ask it to do the research only, then ask it to make the slides!.
0
0
1
@WangMagic_
Wang Magic
1 month
RT @theMetaStoneAI: Say hello to MetaStone-S1!
Tweet media one
Tweet media two
Tweet media three
Tweet media four
0
1
0
@WangMagic_
Wang Magic
1 month
Algorithms and data always move forward in an iterative way.
@_akhaliq
AK
1 month
Scaling Laws for Optimal Data Mixtures
Tweet media one
0
0
1
@WangMagic_
Wang Magic
1 month
MetaStone-S1.5 is on the way….
0
0
3
@WangMagic_
Wang Magic
1 month
RT @_akhaliq: MetaStone-S1 just dropped on Hugging Face. Test-Time Scaling with Reflective Generative Model. Experiments demonstrate that M….
0
52
0