Ryan Feng

1.4K posts

Ryan Feng banner
Ryan Feng

Ryan Feng

@zfeng42

Code & Rock

Katılım Mayıs 2009
378 Takip Edilen145 Takipçiler
Vadim
Vadim@VadimStrizheus·
unpopular opinion: Codex > Claude Code
English
79
7
165
14.9K
Huaizheng Zhang
Huaizheng Zhang@zhzHNN·
Most startups don’t need to fight A-teams in big tech, as there aren’t many A-teams. And now you know, A-teams are not even allowed to ship.
Tibo@thsottiaux

@_chenglou I was part of that team. Basically ChatGPT one year before it came out. Called LMChat and then another codename. Google was too nervous to release it and DeepMind was blocked from shipping products that could disrupt Google. I think about this a lot.

English
3
2
19
2.9K
Ryan Feng
Ryan Feng@zfeng42·
Keep in mind that math is easier to verify than any other science.
English
0
0
0
7
Ryan Feng
Ryan Feng@zfeng42·
@zuess05 This is right. Release before polished will be unacceptable soon.
English
0
0
0
10
Suhas
Suhas@zuess05·
Unpopular opinion: The MVP era is over. People expect your app to actually be good from day one.
English
100
8
393
24.3K
Panta
Panta@thepanta82·
Hermes agent is the worst looking agentic harness or whatever it is I've ever seen. What do people see in this thing?
Panta tweet media
English
78
0
107
38.8K
Kevin S. Xu
Kevin S. Xu@kevinsxu·
Let the record show, as of today, cybersecurity incidents directly instigated by: - a closed model: 5 (Anthropic: 3; OpenAI: 2) - an open weight model: 0
English
17
7
63
6.6K
Ryan Feng
Ryan Feng@zfeng42·
@yetone 我现在只跟Fable谈人生,其他它一无是处
中文
0
0
2
1.8K
yetone
yetone@yetone·
我算是明白了,但凡是涉及虚拟机相关的任务,Claude Code 会自动的把 Fable 5 降级成 Opus 5,这就是为了安全性吗?荒谬荒谬!正值我现在在开发 Agent Sandbox,真的好烦。
中文
27
1
146
36.3K
Kr$na
Kr$na@krishdotdev·
Peak Anthropic customer experience.
Kr$na tweet media
English
29
46
849
24.8K
Victor M
Victor M@victormustar·
ok f*ck it: I deployed a free public endpoint for DeepSeek-V4-Flash-0731, today’s release, on Hugging Face Inference Endpoints. Anyone can use it. No token required. OpenAI-compatible API. It’s a shared community box, so there’s (light) rate limiting. Be nice to your neighbors 🤗 ⬇️ read the guide to start using it with Pi & co
English
63
86
1.1K
165K
DeepSeek
DeepSeek@deepseek_ai·
🚀 DeepSeek-V4-Flash Official API is now LIVE in public beta! 🔷 We’ve massively upgraded its Agent capabilities—benchmark scores are now far surpassing the V4-Pro-Preview. Check out the massive performance leap below! 👇 🔷 The official V4-Flash now natively supports the Responses API format and is fully adapted for Codex! Check out the configuration details in our official API docs: api-docs.deepseek.com/quick_start/ag…
DeepSeek tweet media
English
1.5K
3.4K
28.7K
8.1M
akira
akira@realmcore_·
Everyone who tries to build a harness from scratch inevitably runs into the same set of problems so heres a list of some: 1. Good async execution 2. A resource lifecycle manager 3. Agent harness research/design work 4. Evals 5. etc. (it keeps going) Just fyi
English
38
16
429
52.3K
Ryan Feng
Ryan Feng@zfeng42·
pi and omp are ok tools, and bad products.
English
0
1
0
24
Ryan Feng
Ryan Feng@zfeng42·
Fable 5 has a very deep philosophy core which I don't see in other models.
English
0
0
0
17
soham
soham@soham_btw·
call me 007 0 users 0 products shipped 7 domains bought
English
54
199
3.2K
91.8K