@steipete @_karrtopelka @remwilcox I know, but it’s good to have it. I think it’s worth having a Mac Mini or a GPU to use it.
English
Ernesto
636 posts
English
Who said that openclaw needs Mac mini did not read docs.. you literally can run it on anything. Mine for example runs on Pi and feels good
@steipete am I right?
Nav Toor@heynavtoor
🚨 BREAKING: Someone just rebuilt the entire AI assistant stack in Zig. It's called NullClaw. The binary is 678 KB. It uses ~1 MB of RAM. It boots in under 2 milliseconds. No runtime. No VM. No framework. No garbage collector. Just raw Zig. Here's why this is absurd: → OpenClaw needs a $599 Mac Mini and 1 GB+ RAM → NanoBot needs 100 MB+ RAM and Python → PicoClaw needs 10 MB RAM and Go NullClaw runs on a $5 board with 1 MB of RAM. Same functionality. 0.1% of the resources. Here's what's packed into that 678 KB: → 22+ AI providers (OpenAI, Anthropic, Ollama, DeepSeek, Groq, etc.) → 13 chat channels (Telegram, Discord, Slack, WhatsApp, iMessage, IRC) → 18+ built-in tools → Hybrid vector + keyword memory search → Multi-layer sandboxing (Landlock, Firejail, Docker) → Hardware peripheral support (Arduino, Raspberry Pi, STM32) → MCP, subagents, streaming, voice, the full stack Here's the wildest part: Every subsystem is a vtable interface. Swap any provider, channel, tool, memory backend, or runtime with a config change. Zero code changes. It even encrypts your API keys with ChaCha20-Poly1305 by default. 2,738 tests. ~45,000 lines of Zig. Zero dependencies beyond libc. 100% Open Source. MIT License.
English
English
@remwilcox @steipete I was reacting to post text. Person there said that you NEED Mac Mini, but if you install zig version no need for Mac.. that’s kinda bullshit
English
Apologies, this was a docs clean up we rolled out that’s caused some confusion.
Nothing is changing about how you can use the Agent SDK and MAX subscriptions!
Rob Zolkos@robzolkos
Major Claude Code policy clear up from Anthropic: "Using OAuth tokens obtained through Claude Free, Pro, or Max accounts in any other product, tool, or service — including the Agent SDK — is not permitted"
English
i swear if i see one more “X one shot this” post - that’s what y’all do all day? try to one shot shit? is anyone out there actually using these models for work?
English
@robinebers - Cursor - $20 (Until last month I had Ultra, but I downgraded because I wasn’t using it)
- ChatGPT Pro - $200
- Claude Max 20 - $200
- SuperGrok - $30
- Gemini Pro - $9.99
+ $50 in API usage for my apps (Gemini Flash)
English
here are my active AI subscriptions
cheap if you compare value returned
- Cursor Ultra - $200
- ChatGPT Pro - $200
- Claude Pro - $20
- Grok/X- $40
- Poke - $25
+ $50-100 in API usage for my apps
what about you?
English
Expensive livestream with GPT-5 Pro + Cursor. Does it cook??
Jediah Katz@jediahkatz
GPT-5 Pro is now live in Cursor. It's an extremely deep thinker and good for solving the hardest of problems. It's also quite expensive (on par with Opus) and extremely slow, so I wouldn't use it as your daily driver. To save time, try giving it more context upfront!
English
@RayFernando1337 Not really good at web dev, clearly not on the same level as Claude or Codex. Hard to believe this is Gemini 3.0 Pro
English
@robinebers @nadzi_mouad For me, GPT-5 High (Codex CLI) is still better for everything (design, debugging, etc.), but it’s slow.
Sonnet (CC: Thinking on) feels more dynamic: in tests with existing codebases it felt more solid, faster, and without the bad impression it left before.
English
Zero surprised tbh.
It’s not like Claude models wouldn’t be able to do this. They just don’t reason for as long, and if they did, they’d cost a fortune as the price is still the same.
Very few people will switch back to Sonnet after this release IMO. I’ll test it thoroughly in the next week and then make a big roundup video, but I know the outcome
English
@robinebers Using Cursor just as an editor now. Codex doesn’t really work well there, I think Codex and Claude are better in their own CLIs. I’m 90% on Codex CLI and 10% on CC. If Anthropic doesn’t release something soon, that 10% will drop to 0.
English
DAY 2
Cursor + gpt-5-codex
I spent a good amount today testing this again and I noticed after a couple of hours that I spent more time in the Codex CLI than in Cursor. And that's never a good sign.
I don't think the Codex model works outside of Codex. I don't know why exactly. Maybe it's the way that they trained it. In the next few days, I'm going to go back to GPT-5 in Cursor (specifically the fast variant) to directly compare it with my experience of the last few days.
It's hard to explain why, but gpt-5-codex just feels very slow. It makes a lot of micro changes, which take a lot of time, and it makes mistakes that it wouldn't otherwise make.
Not loving it.
What about you?
English
how can someone affiliated with cline call code-supernova sonnet 4.5?
it's like a mechanic calling a honda a lambo because "well, I heard the engine, lambos have engines, surely it's a lambo?"
Nick@nickbaumann_
it's gotta be sonnet-4.5 right
English
@robinebers Ah, you meant notifications. I was talking about the Following tab on home.
English
@eigarrido How do you do that? I only see tabs for All, Verified or mentions
English
this is either bait or one more person exposed
Chris@chatgpt21
Anthropic, supernova is realllly good at coding
English
@robinebers I don’t even look at the timeline, I only read what the people I follow. If someone starts posting nonsense, I unfollow, that’s what happened here
English
@eigarrido wasn't following him, showed up on my timeline hah
this has turned into such a meme that I think it could be bait as everyone wants to be the one calling others out
English
my mildly hot take is you should unfollow people who say this is sonnet 4.5
Lee Robinson@leerob
There's another new stealth model in Cursor from one of our partners. "code-supernova" is free to try, curious how you find it!
English