I write code 💻 I build things 🚀 I show you the real China | pixel art lover
Developer × AIGC × Content Creator × Entrepreneur
Spicy takes. Real experiences.
The claim and screenshots: https://t.co/fZNMA3JxeJ
(via @ai_funss)
This is user-reported, not an OpenAI announcement. Treat it as a rumor until it's official.
People on the GPT-6.2 Sol gray rollout say it finally dropped the hedge.
No more "I can't confirm" / "this doesn't prove."
It's unofficial. The routing isn't confirmed.
If you think you got Sol, what exact reply made you sure?
The prompt and the original post: https://t.co/pzoNPXW3Oh
(via @delayoxl)
He turned Beck's public ideas into one Opus 5.5 prompt: red-green-refactor, tidy first, and that no-delete-tests rule.
Kent Beck created TDD. Decades later he's using it to keep AI agents honest, and calls it augmented coding, not vibe coding.
I turned his public ideas into one prompt for Opus 5.5: red-green-refactor, tidy first, and a hard rule that it never deletes a test to go green.
Paste it into Claude Code, thank me later ↓
Kent Beck invented TDD. Now he uses it to keep AI agents honest.
He doesn't call it vibe coding. He calls it augmented coding.
The hard rule: the agent never deletes a test just to go green.
Would you paste that into Claude Code?
@HyperAI_News The countdown and the "nO" follow-up: https://t.co/Jzkdj2SZXt
Unofficial early Qwen 4 benches have also been circulating (not the official release). DYOR. Not financial advice.
If Qwen 4 27B drops Nov 9 at 15:00 Beijing time, you'll see it right here 30 minutes later.
If it doesn't, I never said anything.
#qwen#opensource#qwen4
Source, side-by-side video, and the original numbers: https://t.co/n3SWAHPCUG
(via @HeathIsidr64707)
Costs as reported on that post, not a price quote. Not financial advice.
Gemini 4 Carbon (H) vs Claude Opus 5.5 (XHigh) in Blender
Same Blender Python prompt: What will Sisyphus Shark look like in Hungry Shark:World
• Carbon: [1h 29m] · $[8.82]
• Opus 5.5: [1h 10m] · $[15.66]
(only coding)
Which underwater vibe wins for you?
#b3d#blender #Gemini #Claude
Same Blender prompt. Two frontier models. Only the coding bill:
Gemini 4 Carbon: 1h 29m, $8.82
Claude Opus 5.5: 1h 10m, $15.66
Faster cost more. Cheaper took longer.
Which underwater look would you actually ship?
China (Inner Mongolia): ice-and-snow tourism vouchers, 5M RMB pool. 30% off, up to 100 RMB per use, max 10 vouchers, valid 3 days after claiming, through Mar 31, 2027. Claim via UnionPay QuickPass, WeChat mini-apps or Qunar. Not financial advice. DYOR.
https://t.co/YvVhCWvVhi: FLUX 3 Max is 50% off through Nov 21 (site says Nov 21; post says Nov 22). 720p now $0.085/sec. Global. Cheap way to test image/video pipelines before the deadline.
Check out Muse, your personal AI agent. Redeem my code in Settings within 48 hours of joining and we'll both get 1 billion Muse tokens.
Code: PVRESL
https://t.co/FFPjdKo6VB
3/3 Claude Duet (US research preview, invited Pro/Max/Ultra): extra full limit resets (Pro: 3; Max/Ultra: 1), each refilling the 5-hour and weekly caps. Use Settings > Usage on web or Desktop. Unused expire 15 Dec 2026.
https://t.co/5wdTAR1A5R
Not financial advice. DYOR.
1/3 Tencent Cloud TokenHub (CN, real-name required): new-user pack of 1M tokens for language models and 1M for multimodal understanding, valid 1 year from claim. Claim by 31 Dec 2026. Once per main account per model; confirm in the claim popup.
https://t.co/2UmE49MrS9
2/3 Hong Kong Octopus: new or transferred Mobile Octopus users who register and make one HK$100+ spend get HK$150 stored value. 5 Nov 2026–3 Jan 2027 (HK). Limited, first come first served. Tourist app excluded; Android must add it to Google Wallet.
https://t.co/NqBCh0Eubz