Pinned Tweet
I'M UPPING MY P(DOOM)
made a music video where every single frame is Python code. no footage, no editing, no stock images. I directed, Hark Pro by @adcock_brett wrote the code, and 132 versions later it's done.
what's your P(doom)?
#PDoom #CreativeCoding #HarkPro
made a music video where every single frame is Python code. no footage, no editing, no stock images. I directed, Hark Pro by @adcock_brett wrote the code, and 132 versions later it's done.
what's your P(doom)?
#PDoom #CreativeCoding #HarkPro
1
0
0
18
@adcock_brett if anyone's curious how this got made, i kept every version from the first ugly render to the final cut, all 132 of them lol
im-upping-my-pdoom.pages.dev
im-upping-my-pdoom.pages.dev
0
0
0
2
"Rolling out soon" → "Available now."
Gemini 4 Argon is live. Part 2 of my Argon reel, made entirely in code: every frame, sound and note.
1M unified token limit. +35% agentic code at Google. #1 on the Vals Index. $2/$10 per 1M tokens.
#Gemini4 #GeminiArgon
Gemini 4 Argon is live. Part 2 of my Argon reel, made entirely in code: every frame, sound and note.
1M unified token limit. +35% agentic code at Google. #1 on the Vals Index. $2/$10 per 1M tokens.
#Gemini4 #GeminiArgon
While everyone waits on Gemini 4 Argon public release, I had Hark by @adcock_brett turn it into a motion graphics reel. 77.9% on DeepSWE, 68% on CWE-bench tied for #1, autonomous find, validate, patch. The benchmarks say Google is back in the race.
0
0
0
173
My Mac kept saying "storage almost full," so I built my own cleaner.
Meet Reclaim, a native macOS app that shows you exactly what's eating your disk before anything gets deleted. Everything goes to trash first.
Beta testing starts late October. ✉️ DM for early access!
Meet Reclaim, a native macOS app that shows you exactly what's eating your disk before anything gets deleted. Everything goes to trash first.
Beta testing starts late October. ✉️ DM for early access!
1
0
0
59
Missed a pill? Can't remember if you took it?
Petal is a private pill companion whose AI runs on your device (Gemma 3 1B, llama.cpp). No cloud. Works in airplane mode.
Hackathon 2026 · Local AI
#AppBuildersPH @cognition @DevinAI
petal.necookie.dev
Petal is a private pill companion whose AI runs on your device (Gemma 3 1B, llama.cpp). No cloud. Works in airplane mode.
Hackathon 2026 · Local AI
#AppBuildersPH @cognition @DevinAI
petal.necookie.dev
1
1
2
131
While everyone waits on Gemini 4 Argon public release, I had Hark by @adcock_brett turn it into a motion graphics reel. 77.9% on DeepSWE, 68% on CWE-bench tied for #1, autonomous find, validate, patch. The benchmarks say Google is back in the race.
0
0
0
366
People in my previous post on my thread are saying, "old code expired, use mine" 💀
A7YG8R still works. But here's what's actually true:
Your 1 BILLION token bonus does expire if you don't redeem within 48h of joining.
So if you got in with the Gemini trick:
Gemini Pro → pen muse.ai in remote browser" → sign up with Google at muse.ai/join — don't sit on it:
Settings → General → Redeem: A7YG8R
We both get 1B tokens. Don't let it expire.
A7YG8R still works. But here's what's actually true:
Your 1 BILLION token bonus does expire if you don't redeem within 48h of joining.
So if you got in with the Gemini trick:
Gemini Pro → pen muse.ai in remote browser" → sign up with Google at muse.ai/join — don't sit on it:
Settings → General → Redeem: A7YG8R
We both get 1B tokens. Don't let it expire.
1
0
0
72
Stop following VPN tutorials for Muse.
I got in without one in under a minute:
1. Must have: Gemini Pro account
2. Open Gemini Spark Beta
3. Ask it to "open muse.ai in a remote browser"
4. Once it opens a remote browser, click "Take Control" and start signing up in muse.ai (make sure to use a brand new account that isn't yet waitlisted, or this won't work)
No VPN. Done.
Redeem A7YG8R in Settings → General → Redeem within 48h, and we both get 1 BILLION tokens.
I got in without one in under a minute:
1. Must have: Gemini Pro account
2. Open Gemini Spark Beta
3. Ask it to "open muse.ai in a remote browser"
4. Once it opens a remote browser, click "Take Control" and start signing up in muse.ai (make sure to use a brand new account that isn't yet waitlisted, or this won't work)
No VPN. Done.
Redeem A7YG8R in Settings → General → Redeem within 48h, and we both get 1 BILLION tokens.
0
0
0
145
I also now got Muse access using the method below 👇
No US VPN needed (only after successfully getting a working Muse account to access the account in your local device), @berryxia's Gemini trick works:
1. Gemini Pro → Spark Beta
2. Ask it to open muse.ai in a up with Google email: muse.ai/join
We both get 1 BILLION tokens if you redeem my code within 48h:
Settings → General → Redeem invite code: A7YG8R
No US VPN needed (only after successfully getting a working Muse account to access the account in your local device), @berryxia's Gemini trick works:
1. Gemini Pro → Spark Beta
2. Ask it to open muse.ai in a up with Google email: muse.ai/join
We both get 1 BILLION tokens if you redeem my code within 48h:
Settings → General → Redeem invite code: A7YG8R
卧槽，成功开通Muse 不到1分钟。
刚刚刷到老杨说使用Google Gemini 来开通Muse.
赶紧去搞啊，可以送10亿token啊！
【Muse AI开通教程】
试了一下，做了一个视频教程。核心关键：大家首先要准备2个东西。
1. 要有Google的Gemini Pro账号
2. gemini.google.com 登录Pro账号后，点击对应的Spark 话：“帮我使用远程桌面打开Google官网，首页或者之类的这种就都可以。”
输入对应的Muse AI官网，然后使用谷歌邮箱登录，输验证码，就可以了，很简单。
2
0
0
236
holy shit
I just asked @SakanaAILabs Fugu Max:
"can u help me build a single file html threejs game?"
no spec, no genre, nothing
it just BUILT me a full playable game
try it now 👉🏻 chat.sakana.ai
"Free for all to use" - @SakanaAILabs
I just asked @SakanaAILabs Fugu Max:
"can u help me build a single file html threejs game?"
no spec, no genre, nothing
it just BUILT me a full playable game
try it now 👉🏻 chat.sakana.ai
"Free for all to use" - @SakanaAILabs
🐙 Sakana Fugu Max is now on Sakana Chat 🐙
Free for all to use.
Try it: chat.sakana.ai
Blog: sakana.ai/chat-fugumax/#… 🐡
0
0
0
81
Jnvhr retweeted
🎁 We’re giving away 1,000,000,000 tokens.
Terms and conditions:
✅ Follow @APInex_ai
🔁 Repost this post
💬 In the comments: your favourite AI model + tag 2 friends
10 winners. 100 million tokens each — credited to your APInex balance.
Results on 14 September. 10 winners selected at random.
Terms and conditions:
✅ Follow @APInex_ai
🔁 Repost this post
💬 In the comments: your favourite AI model + tag 2 friends
10 winners. 100 million tokens each — credited to your APInex balance.
Results on 14 September. 10 winners selected at random.
47
40
27
2,398
Jnvhr retweeted
6 months claude pro (5x & 20x plans) on a steep discount today😳
offer details:
tier 1: 5x Plan (6 Months)
tier 2: 20x Plan (6 Months)
discord link :[discord.com/invite/Du8HzAv…]
reddit-exclusive, so it drops straight into a discord ticket
how to claim it:
-repost this post
-comt (5x or 20x)
-join Discord and open a ticket with your Reddit username
inventory i also saw in their slots (i checked it just now):
→5x Plan (6 Months) - 5 coupon slots left
→20x Plan (6 Months) - 4 coupon slots left
third-party discount, not the official Anthropic billing page. limited slots can run out without notice, so jump in early to lock your discounted price
save this if you wanted 6 months of Claude Pro without paying full price
offer details:
tier 1: 5x Plan (6 Months)
tier 2: 20x Plan (6 Months)
discord link :[discord.com/invite/Du8HzAv…]
reddit-exclusive, so it drops straight into a discord ticket
how to claim it:
-repost this post
-comt (5x or 20x)
-join Discord and open a ticket with your Reddit username
inventory i also saw in their slots (i checked it just now):
→5x Plan (6 Months) - 5 coupon slots left
→20x Plan (6 Months) - 4 coupon slots left
third-party discount, not the official Anthropic billing page. limited slots can run out without notice, so jump in early to lock your discounted price
save this if you wanted 6 months of Claude Pro without paying full price
cost at $0.75/$3.75 per million tokens, Gemini 3.7 Flash might be the best price-to-performance ratio in Google's lineup right now😳
what it's genuinely good at:
-coding agents →78% SWE-bench, handles multi-step debugging and repo-level changes
-multimodal→ image, PDF, video, oss all of them
-long context→ 1M tokens, dump a 700-page doc and ask real questions
-iterative dev loops→ fast and cheap enough to run all day
where it struggles:
flagship creative writing, deep research chains, niche domains without fine-tuning
the math is simple:
$0.75/$3.75 per million tokens and it just works for most real workflows
API docs: Gemini 3.7 Flash | Gemini API | Google AI for Developers
72
51
118
40,235
🚨 HUGE UPDATE: 2 new heavyweights just landed on Agent Router!
Now live on single API endpoint:
⚡ @ZhipuAI GLM-5.3 (glm-5.3)
⚡ @DeepSeekAI V4 Flash (deepseek-v4-flash)
Route seamlessly between bleeding-edge speed & reasoning without juggling multiple API keys:
💎 @Ant-8 | claude-opus-5
🔥 @OpenAI: gpt-5.6-sol
⚡ @deepseek_ai: deepseek-v4-flash
🧠 @ZhipuAI: glm-5.3
1 Token. 5 Flagship Models. Zero Downtime.
👉 Test them now in your own harness [mnkr.pages.dev/agentrouter]
#AI #AgentRouter #DeepSeekV4Flash #GLM53 #ClaudeOpus #OpenAI
Now live on single API endpoint:
⚡ @ZhipuAI GLM-5.3 (glm-5.3)
⚡ @DeepSeekAI V4 Flash (deepseek-v4-flash)
Route seamlessly between bleeding-edge speed & reasoning without juggling multiple API keys:
💎 @Ant-8 | claude-opus-5
🔥 @OpenAI: gpt-5.6-sol
⚡ @deepseek_ai: deepseek-v4-flash
🧠 @ZhipuAI: glm-5.3
1 Token. 5 Flagship Models. Zero Downtime.
👉 Test them now in your own harness [mnkr.pages.dev/agentrouter]
#AI #AgentRouter #DeepSeekV4Flash #GLM53 #ClaudeOpus #OpenAI
0
0
1
80
You can currently get ~$150 in FREE API credits on Agent Router to test 3 frontier models 🤯
Here's what you can get:
🎁 ~$100 starter credits just for signing up
🎁 + $50 EXTRA bonus credits if you use my invite link👉 Total: ~$150 in free credits
Models available right now:
in Claude Code CLI - fast streams)
• claude-opus-4-8
• gpt-5.6-sol
Plus 0.2x prompt caching discount built in.
Claim your ~$150 credit stack here 👇
agentrouter.org/register?aff=6…
#ClaudeCode #AI #Anthropic #OpenAI #LLM #DevCommunity #Coding
Here's what you can get:
🎁 ~$100 starter credits just for signing up
🎁 + $50 EXTRA bonus credits if you use my invite link👉 Total: ~$150 in free credits
Models available right now:
in Claude Code CLI - fast streams)
• claude-opus-4-8
• gpt-5.6-sol
Plus 0.2x prompt caching discount built in.
Claim your ~$150 credit stack here 👇
agentrouter.org/register?aff=6…
#ClaudeCode #AI #Anthropic #OpenAI #LLM #DevCommunity #Coding
0
0
0
64
How to route claude-opus-5-thinking to Cursor / Cline / IDEs with 0 proxy setup:
• Base URL: gorouter.app/v1
• Format: Chat completions
• Model: claude-opus-5-thinking
Full OpenAI SDK compatibility for code gen.
Free starter credits in reply 👇
• Base URL: gorouter.app/v1
• Format: Chat completions
• Model: claude-opus-5-thinking
Full OpenAI SDK compatibility for code gen.
Free starter credits in reply 👇
1
0
0
40
Claim your starter credits here:
🔗 gorouter.app/sign-up?aff=p6…
• Works with: Cursor, Cline, ZCode, any IDEs
• Privacy: Zero prompt logging / OpenAI-compatible BYOK
Drop a comment if you need custom config files or latency benchmarks.
🔗 gorouter.app/sign-up?aff=p6…
• Works with: Cursor, Cline, ZCode, any IDEs
• Privacy: Zero prompt logging / OpenAI-compatible BYOK
Drop a comment if you need custom config files or latency benchmarks.
1
0
0
12
🛠 Cursor / Cline / any IDE settings.json snippet:
{
"apiProvider": "openai-compatible",
"baseUrl": "gorouter.app/v1",
"apiKey": "YOUR_KEY",
"customModelId": "claude-opus-5-thinking"
}
💡 Supports reasoning stream & auto-fallback.
{
"apiProvider": "openai-compatible",
"baseUrl": "gorouter.app/v1",
"apiKey": "YOUR_KEY",
"customModelId": "claude-opus-5-thinking"
}
💡 Supports reasoning stream & auto-fallback.
0
0
0
3