Tokenstead

Your local AI Community

[Tokenstead @tokenstead]

@TheAhmadOsman True, the value prop is not really that strong and reselling a pimped out Apple is likely to retain value longer and better than a DGX will...

[Tokenstead @tokenstead]

@theo I just found adding logo to project folder, love this feature

[Tokenstead @tokenstead]

@tonydehnke @rails @OpenAI @Alibaba_Qwen @AnthropicAI Awesome idea Tony, I might add this to our site

[Tokenstead @tokenstead]

After just trying out T3 Code I can say that it got the DevX right. I can actually see markdown files (like full preview), I can have organized projects and agents inside all running whatever harness/model I want.
Running it right now with OpenCode and it has been fantastic.

[Theo - t3.gg @theo]

Wait what

[Tokenstead @tokenstead]

🧐

[OrcaRouter 🐳 @OrcaRouter]

We’re taking uncensored model APIs off general public access.

We built them for research and security. Some users abused them for unlawful purposes. That crosses the line.

Going forward: KYC/KYB-verified companies and legitimate security research only.

Regular Qwen3.8 (model name: qwen/qwen3.8-27b) stays open: $0.33 input / $2.40 output per 1M tokens https://t.co/SD4GVlJxIR

Our mission remains the same: make AI infrastructure and security open, secure, and accessible. Security shouldn’t be a privilege reserved for the largest companies.

OrcaCyber launches next month. 🛡️
OrcaRouter users will be first.

Power to defenders. 🐳

[Tokenstead @tokenstead]

Is it just me or does OpenCode keep going in these insane loops for no reason? Is there an update to resolve this? Basically unusable rn

[Tokenstead @tokenstead]

@MichaelGannotti 😍😍😍
Looks wicked!

[Tokenstead @tokenstead]

@finelytunedai This is the way!

[Tokenstead @tokenstead]

@Sakshi50038 Vibe coders be like

[Tokenstead @tokenstead]

@0xNeoArch One of the best Rig names yet!

[Tokenstead @tokenstead]

The most cost effective option, for cloud, is Opencode Go with Deepseek V4 Flash. It's a workhorse, highly capable model and generous usage capacity. I have 3 terminals going for the last few hours, still no cap limits.

[Tokenstead @tokenstead]

@LexmountAI Interesting, does this replace the default browser in most harnesses or is that required through skills?

[Tokenstead @tokenstead]

How are open-weight models in comparison, when asking these questions?
Kimi, Deepseek, GLM, Qwen

[Mary Talley Bowden MD @MaryBowdenMD]

Be very careful about using AI for healthcare questions. Grok misdiagnosed my MRI, still thinks COVID shots are safe and effective and will argue that Hydroxychloroquine/ivermectin don’t work.

[Tokenstead @tokenstead]

@BlumeDotCodes @olavlj @pederaaby Added to our index, new category
https://t.co/klkgkwDaYL

[Tokenstead @tokenstead]

@omkarships https://t.co/obcUWoXKw5

[Tokenstead @tokenstead]

@BlumeDotCodes Can it be added to support OpenCode harness?

[Tokenstead @tokenstead]

@BlumeDotCodes Interesting, so it's monitoring all real-time harness input/output or is this post completion work that it evaluates?

[Tokenstead @tokenstead]

@theo Call it orchestration, harness control plane?

[Tokenstead @tokenstead]

@TheAhmadOsman @NVIDIAAI and GLM 5.3!

[Tokenstead @tokenstead]

Somehow I cancelled my Ollama MAX plan by accident and now I'm basically model providerless. ☹️
After poking around trying to see awesome cost/benefit, I decided to give @opencode Go a try using Deepseek V4 Flash. Granted this is super cost effective $5/mo for first mo, I will likely max it out in 30 min from now. Will report back!