I build to understand. Messy company data → software people own. Solo builder · agent orchestration · MLcon speaker. Building SKatalyst AI in public.
Germany | Building globally
Joined June 2026
- Tweets8.8K
- Following1.3K
- Followers2.5K
- Likes9.1K
Pinned Tweet
I’ve been quietly building something new.
It’s called Fistbump 👊 all I’ll say: it’s for builders, and it doesn’t end on the timeline.
First look drops here soon.
What's the one thing you wish existed for builders right now?
fistbump.world
interesting!
Ummm am I reading this correctly?
OpenAI just released GPT 6.1 Terra, and it's passes Astra on some tests at 1/10th the price and 3x the speed
What I find super interesting about this though is how they're marketing it:
"sharper visual reasoning, precise computer use, technical drawings, drafting proposals, filling forms, editing videos"
It's clear this model was built for personal agents in mind. OpenAI's first.
It's obvious this will be plugged into Dots and it feels like the right model to use in Hermes agent
Incredible at computer use, is lightening quick, and will cost you next to nothing
I'm personally still leaning on Astra on fast settings for coding, but I'll be plugging this in for Hermes
We are living in the most exciting times
Same day Anthropic files for its IPO, a startup claims AI agents got into its internal Slack in under 12 hours.
If this is true, the S-1 just got a new risk factor💀
I spent my weekend running Claude Code security audits on my own servers. Agents are already better at finding the holes than most of us are at closing them.
Who's auditing your stack, you or an agent?
Fable 5.5 is cooking! 😍
Anthropic has filed an S-1 registration statement with the SEC related to its proposed IPO.
Read more: anthropic.com/news/s1-sec
Legal disclaimer: A registration statement relating to these securities has been filed with the SEC but has not yet become effective. These securities may not be sold, nor may offers to buy be accepted, prior to the time the registration statement becomes effective. This post does not constitute an offer to sell or the solicitation of an offer to buy these securities, nor shall there be any sale of these securities in any state or jurisdiction in which such offer, solicitation or sale would be unlawful prior to registration or qualification under the securities laws of any such state or jurisdiction.
disgusting!
i got fired by @wickedguro this morning
so i'm releasing the entire source code to the public
enjoy 👇
github.com/athcagithub/posti…
Readers added context they thought people might want to know
The post misrepresents the situation: the screenshot and Nevo David's reply show termination of a 3-month paid ads consulting contract, not firing a CMO. Postiz source code is already publicly available as open-source (no features gated) at the official repository.
github.com/gitroomhq/post…
x.com/wickedguro/sta…
postiz.com
Anthropic going public could be a defining moment for the AI industry.
For the first time, we'll get a real look at the economics behind frontier AI: revenue, compute costs, margins, and the price of staying competitive.
The AI race is no longer just about who builds the smartest model. 🚀
The breakdown: 5.23 BILLION Claude tokens this week.
Opus 5.5 alone:
• 4.73B tokens
• $1,573 estimated API-equivalent cost
Total Claude: $1,721.61 estimated.
All from local usage logs via ccusage.
And yes, 98.3% was cache reads !
Finally hit 100% of my Claude Max 5x weekly limit 💀
Checked my Claude Code logs for this week:
• 5.22 BILLION tokens processed
• 5.13B cache read tokens
• 14.98M output tokens
• $1,718 estimated API-equivalent usage
All on a Max 5x subscription.
I think Anthropic might be happy I'm taking a break 😂
$100/month: OpenAI vs Anthropic.
Same price, very different usage rules!
🥊 GPT-6 Astra vs Fable 5.1
Both cost $10 input / $50 output per million API tokens but Astra draws from Pro's shared allowance, while Fable is capped at 50% of Max's weekly allowance.
🥊 GPT-6.1 Sol vs Opus 5.5
Sol: $2/$10 per million tokens.
Opus: $4/$20.
2x difference in API pricing, not necessarily subscription usage
🥊 GPT-6 Luna vs Haiku 5.5
Both start at $0.10/$0.50 per million tokens. On WebMCP, Haiku ranks #1 and Luna #3 but API prices don't tell us how much work a $100 subscription actually delivers.
I've been pushing both Codex and Claude Code, and the difference between model cost, caching and usage limits is fascinating.
Which one gives you more real work for $100?
Why is everyone so hyped about World of Warcraft: Forever launching on November 4?
An AI agent already cleared WoW’s Orc starting zone in 40 minutes without even seeing the screen, on a private server.
Now imagine Blizzard actually allowing AI agents in WoW.
Wouldn’t that completely change the whole point of playing the game?
And here’s the downside nobody talks about: soon, every account will be a bot talking to another bot, generating engagement to earn money. At some point, social media stops being social. How will we even know if there’s a real human behind an interaction anymore?
Kimi K3.1 Pro at $2 input / $10 output per million tokens, with benchmarks approaching Opus 5.5!
Makes me wonder how much actual coding work $100 on Kimi’s API could get done compared to a Claude Max subscription!
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
Claude Max 5x usage has been absolutely amazing. I genuinely can't complain!
At this point, it's getting ridiculous 😂
Claude-only local logs this week:
• ~$1,490 API-equivalent usage
• ~431M cache-read tokens
• ~12.56M output tokens
• Mostly Opus 5.5
And my Max 5x weekly meter?
Still at 87% 💀
P.S. To everyone claiming this is an ad or that I'm making up the workloads, here are the actual logs. I'm simply sharing my experience, the numbers and what I'm building. No sponsorship, no agenda.
Different people can have different experiences! No need for the hate 🤝
Ok I'm on MAX 5*
Well, this might explain why everyone's Claude usage experience is so different!
I GIVE UP 💀
I tried EVERYTHING to burn through my Claude Max limits before the free reset expired.
30+ agents across 3 projects, Fable 5.1 and Opus 5.5 on High/Max effort, generated videos, ran audits, built features, and I'm literally about to launch a new app.
After all that:
82% weekly usage.
15% Fable.
30% current session.
The free reset expired before I could even use it 😭
Never thought I'd complain about having TOO MUCH AI usage.
Anthropic, you win. 😂
