buzz 1.0 is finally here! the most flexible team workspace out there. for humans and agents, model agnostic, local or cloud, and infinitely extendable.
buzz.xyz
Salman Mohammed retweeted
This is it, everyone!
We released Buzz 1.0, and what an update it is.
Your people, your agents, your project, all in one place.
Plugins. Make Buzz work your way.
Plugins are the heart of Buzz. Add your own tools, pages or panels, extend conversations, and build the workspace you want. Share with your team or with the community.
Work solo. Work together.
Your own agents chats and workspaces alongside team channels. Bring others into a chat when your're ready to collaborate.
Your choice of agents.
Run Claude, Codex, Goos or Pi on your computer. Create teams of agents for different kinds of work.
Go beyond typing.
Send voice notes, jump into live audio Huddles with yuour team, or talk with supported voice-capable agents.
Ask your agents to keep and eye on things.
With attention, they can watch for updates, follow a PR's checks and reviews, and report back while buzz is running.
On top of that we released a new version of shared compute, a plugins marketplace we will be opening up soon, agents marketplace and lots more to come.
Download the new version from buzz.xyz
Enjoy!
my usage over the past month: went from 90% gpt 5.6 sol to 75% opus 5.5 (on buzz & claude desktop app)
gpt 6.1 sol is really good, but opus 5.5 feels better overall, for both design & code.
two reminders:
1. competition is a beautiful thing
2. model >>> harness
89
a weird inversion with LLMs is the models improve faster than the tinkerers
when i see people with custom workflows and setups they're all addressing problems that don't exist anymore
the person naively using vanilla codex is more likely to be experiencing state of the art
Salman Mohammed retweeted
"Harness choice has little effect on task success rate ... As models become more capable, agents may need less scaffolding."
Does your Claude model really need Claude Code…? 🤔
We evaluate 7 models on Claude Code, Codex, and Pi. Three surprising findings emerge:
1️⃣Harness choice has little effect on task success rate, but can significantly affect the cost
2️⃣A simple harness can be competitive
3️⃣The native harness isn’t always the best.
Millions of people are using coding agents, but the impact of harness choice remains unclear.
(1/n) More details in the thread. 🧵
Salman Mohammed retweeted
Not enough people are seriously trying to reconcile the two facts:
- models can now resolve Millennium Problems
- annualized US GDP growth is at 1.5%
Salman Mohammed retweeted
The narrative around why building AGI was a good idea has always been about invention -- "it will cure cancer", "it will solve fusion energy", etc.
So I would not "declare AGI" until we have AI that is, at last, capable of invention -- conceptual breakthroughs, novel insights, new real-world technology, etc.
aligns with my personal experience:
- codex >>> claude code or desktop app
- pi is really good esp at cost / task & i wanna avoid lock-in
my current tools:
- codex desktop app for single player
- pi in the terminal
- buzz for collaborative work
A year ago the question was which model. Now it's which harness.
Pi, Exo, Claude Code, Codex, DeepSeek Harness and 4 others. Same model, same tasks, same runtime. 360 runs, 2 billion tokens.
Pass rates: 50% to 67%.
Cost per pass: $1.05 to $18.34.
Introducing FrontierHarness Eval. 🧵
1
1
146
Salman Mohammed retweeted
I've been thinking about this quote for a solid 24 hours
It's a like a raspberry ripple of bullshit
A vanilla ice cream of VC funding, stirred with a little pinch of turd
Salman Mohammed retweeted
Its crazy how the consensus shifted away from Claude being the favorite to Codex
It’s not just Vibes. Things favoring codex:
- Better harness (Codex Desktop)
- Better pricing: more things included within subscription (fast mode etc). bonus - constant resets from Tibo 😅
- Less refusals / false positive safety triggers
- You can use subscription freely wherever you want (Opencode, Amp, etc)
When you add all these up it makes codex significantly more attractive. There is likely an industry lag of people still using Claude but feels like it’s gonna swing back
Salman Mohammed retweeted
Glad to see this -- renderers are a foundational component of the LLM stack. Renderers map between tokens and messages, which are invariant to tokenizer and formatting details. Most APIs, datasets, and RL environments are defined in terms of messages.
Getting the details wrong leads to train-test mismatches, caching inefficiencies, and prompt injection vulnerabilities. We included a renderers module in Tinker Cookbook, but it makes sense as a standalone library.
deepseek v4 pro & flash work incredibly well with coding agents! this should put a ceiling on how much anthropic / openai can raise subscription prices.
129
Salman Mohammed retweeted
I strongly believe there are entire companies right now under heavy AI psychosis and its impossible to have rational conversations about it with them. I can't name any specific people because they include personal friends I deeply respect, but I worry about how this plays out.
I lived through the great MTBF vs MTTR (mean-time-between-failure vs. mean-time-to-recovery) reckoning of infrastructure during the transition to cloud and cloud automation. All those arguments are rearing their ugly heads again but now its... the whole software development industry (maybe the whole world, really).
It's frightening, because the psychosis folks operate under an almost absolute "MTTR is all you need" mentality: "its fine to ship bugs because the agents will fix them so quickly and at a scale humans can't do!" We learned in infrastructure that MTTR is great but you can't yeet resilient systems entirely.
The main issue is I don't even know how to bring this up to people I know personally, because bringing this topic up leads to immediately dismissals like "no no, it has full test coverage" or "bug reports are going down" or something, which just don't paint the whole picture.
We already learned this lesson once in infrastructure: you can automate yourself into a very resilient catastrophe machine. Systems can appear healthy by local metrics while globally becoming incomprehensible. Bug reports can go down while latent risk explodes. Test coverage can rise while semantic understanding falls. Changes happens so fast that nobody notices the underlying architecture decaying.
I worry.
Salman Mohammed retweeted
this is so dumb ...
arstechnica.com/ai/2026/05/a…
bad take imo. i'd prefer to use my custom shitty harness with opus 4.7 than claude code with 3.7. model matters wayyy more than the harness today. simplicity of pi is a great example.
.@addyosmani shares hot takes:
"A decent model with a great harness beats a great model with a bad harness"
"The gap between what today’s models can do and what you see them doing is largely a harness gap."
"A harness is a living system, not a config file you set up once"
addyosmani.com/blog/agent-ha…
1
104
doesn't surprise me this is true. the safety instructions in system prompts take up 1000s of tokens & it's usually irrelevant to the user's ask
teapawt.net/i/status/2046304921134…
opus 4.7 seems to have a much better time in claude code if you run without most of the system prompt (claude --system-prompt ".")
1
36























