YSY

Blessed for meeting amazing, inspiring and loving people. Love to pair technological innovation to human needs.

[YSY @edu_yeh]

Speed is the least interesting thing about working with AI agents.
What's interesting is what happens when thought no longer has to wait to become form — and what that asks of the human side of the collaboration.
New essay. ↓

[YSY @edu_yeh]

https://t.co/R5zWpg3fUL

[YSY @edu_yeh]

@thsottiaux 🙏🙏🙏🙏🍾🍾🍾🍾

[YSY @edu_yeh]

@theo Me too 😢

[0xSero @0xSero]

New American open weight model, 950B params, huge one. All modalities in.

Possibly the coolest demo I’ve ever seen

[YSY @edu_yeh]

@tibo_maker Where? Where ? My $200 plan reached max and have to wait about a week 😢😢😢😢😢

[Daniel Cukier @danicuki]

YouTube Channel Revamper
Revamp an entire YouTube channel — thumbnails, titles, and descriptions — end to end.

I just created this project for myself, and maybe this can be helpful for others.

https://t.co/tBEFw8SFMg

It pulls your whole upload history, reads each video's transcript, asks a fast LLM for a stronger title, a thumbnail headline, and a rewritten description with real chapters, generates an on-brand image with Gemini, composites it into your brand template, and lets you review everything in a local UI — sorted lowest-views-first, where the upside is biggest.

[YSY @edu_yeh]

@thsottiaux Congratulations and we really appreciate the removal of some of the restrictions so we can continue to test and collaborate with these powerful models

[YSY @edu_yeh]

@wholemars 😂😂😂😂😵‍💫

[Tibo @thsottiaux]

Thank you to the 7M active users who are now using Codex and ChatGPT Work.

We have added a banked reset to everyone's account to celebrate the milestone. You can apply the reset in the desktop app or on web and it will replenish the weekly usage for you.

Have fun out there.

[YSY @edu_yeh]

@satyanadella This should not just be applicable to Enterprises but individuals

[Aryan Butala @butala_aryan]

CRITICAL FSD disengagement at highway speed.

Details:
~ car was extremely indecisive taking the exit
~ You can see right before I disengaged, the steering wheel was turning slightly left toward the barrier
~ also waited until the very last minute, cutting across 4 lanes to make the exit in under 10 seconds
~ I believe it was going to drive into the barrier
~ Mad Max mode
~ The car behind me was cut also about to collide due to this indecisiveness
~ FSD v14.3.4 on 2026.14.6.12
~ Completely sunny day, sun was above not in frony (no glare)
~ All cameras were clean
~ Houston

I am a very grateful and avid FSD user, but never in my 3 years of usage have I experienced such a critical disengagement. I made sure to mark it as critical in the menu. Also, the video clip doesn't really justify how much I had to swerve and avoid that barrier, also having to accelerate hard since the car behind me didn't expect that.

Please repost and share so @Tesla_AI can see this and fix it 🙏
@elonmusk

[Tibo @thsottiaux]

@mylifcc This is not correct. Do not do this if you do not understand exactly what you are doing.

We do not charge extra above 270k context and the context threshold has been tuned for GPT 5.6 Sol to be perfect at the default limit.

[YSY @edu_yeh]

@mattshumer_ Sorry to hear that! I will make sure that your pain won’t go wasted by implementing a prevention against such unfortunate and tragic incident causes by agents

[YSY @edu_yeh]

As usual, upon an update, many of the connectors they were already installed and up and running, stopped working I had to spend significant time to reset, and ensure that the environment was back in working conditions again. This is always one of the most painful experiences when dealing with major upgrades wish there was a better way where the agents would first evaluate the current settings which involves all the plugins and connectors and then ensure that there are still working after the migration gets completed

[Vincent @Vincent_AINotes]

GPT-5.5 Codex 变笨的真相找到了。

44%的复杂任务会在516个token被强制截断，错误率飙升至83%。

不是模型降级，是OpenAI服务层的调度机制把它"掐断"了。

完整分析和3个一行配置就能解决的方案👇

[Claude @claudeai]

We're extending access to Claude Fable 5 on all paid plans through July 12.

[YSY @edu_yeh]

@aimikoda @Korgamiai And how accurate and controllable it is using mannequin storyboard compared to other methods.

[Kinokuniya USA @KinokuniyaUSA]

Kinokuniya Los Angeles in Little Tokyo will begin its soft opening today at 4pm! 📢📚✏️📓

Thank you all for your patience as we relocated our Los Angeles store, now the second-largest Kinokuniya store in the U.S.!

Enjoy more room to browse with wider aisles, higher ceilings, and an expanded selection of books, stationery, Japanese gifts, and more. Stop by to rediscover your favorites or experience our new store for the first time—we can’t wait to welcome you!

Address:
101 San Pedro Street
Los Angeles, CA 90012

Hours:
11:00am - 8:00pm (Daily)

[YSY @edu_yeh]

@latvianotter @thsottiaux I got the reset ! Thank you @OpenAI

[Ornith @ornith_]

Aloha! 🌺 Meet Ornith-1.0, a family of open-source LLMs specialized for agentic coding.

Ornith-1.0 spans the full parameter sizes including 9B Dense, 31B Dense, 35B MoE, and 397B MoE. It achieves state-of-the-art performance among open-source models of comparable size on coding benchmarks including:
✅Terminal-Bench 2.1(77.5)
✅SWE-Bench(82.4 on verified, 62.2 on pro, 78.9 on Multilingual)
✅NL2Repo(48.2)
✅SWE Atlas(41.2 on QnA, 42.6 RF, 39.1 TW)
✅ClawEval(77.1)

Post-trained on top of gemma4 and qwen3.5, Ornith-1.0 employs a novel self-improving training strategy in which reinforcement learning is used to generate not only solution rollouts, but also the task-specific scaffolds that drive those rollouts. By jointly optimizing the scaffold and the resulting solution, the model generate higher-quality solutions in agentic coding.😎

All models are released under the MIT license, enabling full commercial and research use.

📖Tech Blog: https://t.co/qT9N2HYWFn
🤗Huggingface: https://t.co/PRrwqjeBtM