khushal

im mostly sleeping or staring at my screen :|

[khushal @celeroncoder]

@Alibaba_Qwen @opencode when is this coming ?

[khushal @celeroncoder]

@nucleusagi @rvaniaaaa a lot of 1M models just came out, 1 million context windows sounds like a lot but most models loose the context after reaching just 10-20% of it. That's why you need a persistent context layer

[khushal @celeroncoder]

@robinebers how did you set that up?

[khushal @celeroncoder]

this was the inspiration btw https://t.co/bbUnv2metQ

[Toby @TobyBelhome]

Real estate detail page with shadcn/ui & tailwind css.

⤵️ The link is below.

[khushal @celeroncoder]

m3 one shotted this with a single prompt

[MiniMax (official) @MiniMax_AI]

Introducing MiniMax M3: The First Open-Weights Model to Combine Three Frontier Capabilities

- Coding & Agentic Frontier: 59.0% SWE-Bench Pro, 66.0% Terminal Bench 2.1, 34.8% SWE-fficiency, 28.8% KernelBench Hard, 74.2% MCP Atlas
- MiniMax Sparse Attention scales context to 1M
- Natively Multimodal from Step Zero

API: https://t.co/fHRdSV7BwZ
Token Plan: https://t.co/BDCycxepZw
🚀New! MiniMax Code: https://t.co/GvB4YiB6Ul

Weights & Tech Report in ~10 Days

[khushal @celeroncoder]

@robertcourson how do i try this ?

[khushal @celeroncoder]

used this feels like somewhere b/w opus and gpt 5.5

[JUMPERZ @jumperz]

Minimax m3 is wild and it broke the one rule every ai model has followed which is better costs = better capability...

if you put every model on a graph, price on one side, how good it is on the other.. they all fall along a straight line..

cheap / weaker models sit bottom left and expensive / stronger ones sit top right.. you pay more, you get more simple as that..

picture a diagonal from cheap and weak (bottom left) to expensive and strong (top right)..

that line is the going rate of how much capability your money buys... every model pays it... when m3 is the first to get more than it paid for, landing above the line where nothing has ever been..

it's as capable as the mid tier frontier models, but priced like the cheapest ones $1.20..

and the bigger part is that m3 is open weight so for the first time, the best value on the chart is also the one you fully own..

[JUMPERZ @jumperz]

Minimax m3 is wild and it broke the one rule every ai model has followed which is better costs = better capability...

if you put every model on a graph, price on one side, how good it is on the other.. they all fall along a straight line..

cheap / weaker models sit bottom left and expensive / stronger ones sit top right.. you pay more, you get more simple as that..

picture a diagonal from cheap and weak (bottom left) to expensive and strong (top right)..

that line is the going rate of how much capability your money buys... every model pays it... when m3 is the first to get more than it paid for, landing above the line where nothing has ever been..

it's as capable as the mid tier frontier models, but priced like the cheapest ones $1.20..

and the bigger part is that m3 is open weight so for the first time, the best value on the chart is also the one you fully own..

[The AI Brain Company / Nucleus AI @nucleusagi]

We launched World Context on Day 2 of the US-Israeli strikes on Iran.

The headlines were already wrong.

30 days later the system has autonomously built:

4.7 million coherence links 298,000 verified claims.

190,000 tracked events.

Zero human curation.

Aggregation was never the hard part. Revelation is.

[The AI Brain Company / Nucleus AI @nucleusagi]

One conversation in Claude. One query in ChatGPT. Zero copy-paste.

This is what the context layer does. Your models don't need to talk to each other. They just need access to the same brain.

[The AI Brain Company / Nucleus AI @nucleusagi]

150 articles. 20 years of geopolitical history.

One query.

This is what happens when you give an AI model a real context layer instead of relying on training data alone.

[khushal @celeroncoder]

@Raakin added all those articles to claude using MCP only!

[khushal @celeroncoder]

GPT is great at creative writing (ps i love talking to it); Claude reasons well but you need to move context between them.

Use Nucleus AI, it solves "the context problem", drop it once and you can use it anywhere.

dm/comment to skip the waitlist

[The AI Brain Company / Nucleus AI @nucleusagi]

Your AI has amnesia.

Every time you switch between Claude and ChatGPT, you start from zero. You copy. You paste. You re-explain your entire business to a supercomputer that forgot you existed.

We got tired of it. So we built the fix.

Nucleus AI Brain is a memory layer that sits across your AI tools. Save your best outputs. Keep your context intact. Move between models without losing your train of thought.

AI models change. Your context remains.

We've got founding members already inside. Today we're opening up priority waitlist spots.

Reply with BRAIN and we'll DM you the link to skip the line.

[khushal @celeroncoder]

@theo If anthropic was a good company we'd have claude code too

[khushal @celeroncoder]

@sunithvs_ @supabase I am still getting the HSTS error on @WisprFlow supabase auth

[Mike Rundle @flyosity]

--dangerously-skip-permissions

[khushal @celeroncoder]

wispr flow couldn;'t sign in i thought something broke so tried downloading again but now can't login

because @WisprFlow requires signin to even download and that broke because supabase is broken in India

i hate this

[khushal @celeroncoder]

@nexxeln gpt 5.2 got way smarter

[khushal @celeroncoder]

opened claude code inside 1code terminal because it does not reconizes my custom commands :/

@serafimcloud

[khushal @celeroncoder]

@brian_lovin try out https://t.co/9VEph8CJYq https://t.co/P45pkZPOMs

[Charlie Lamb @charlietlamb]

Introducing Ferix - the skill issue killer.

Ferix is built around @vercel's ‘npx skills’ command which @rauchg announced yesterday.

Skills are stored in repositories on GitHub which means finding the skills you need for your project isn’t intuitive and you have to go searching. Ferix fixes this problem and acts as a central hub for all skill repositories.

You an add a repository in 5 seconds and easily view it’s contents. So far I’ve added @better_auth, @expo & @vercel's agent skills but any user can add a supported repository.

Let me know what you think and anything that can be improved!

@autumnpricing skills coming soon👀