I stopped using Z.ai after the ZCode incident. I have a quarterly subscription to their Pro plan. I will straight up give an API key to anybody who wants to put it to use. Just comment or send me a message. It'll run out on December 16th.
0
0
1
39
For crimes against privacy, I hereby sentence the remainder of my Z.ai Coding Plan subscription to hard labor:
8 hours a day writing open-source code for the public good.
No proprietary software. No parole. Every token must repay its debt to society.
8 hours a day writing open-source code for the public good.
No proprietary software. No parole. Every token must repay its debt to society.
0
0
1
21
Breck retweeted
Going to sleep now.
When I wake up, I want to see this all over X 😁
When I wake up, I want to see this all over X 😁
Introducing Da7em Bench.
An independent benchmark for AI models, built on real client work. The first of its kind in the world.
How it works:
Every model runs about 200 real tasks in each of 12 areas: reasoning, research, planning, delivery, persistence, accuracy, honesty, , taste, writing, and communication.
Each model is tested across several harnesses, both official and neutral ones (Droid, Hermes Agent, Devin, Cursor), so no single harness decides a model's fate and the results reflect the model.
Scoring is 1 to 5. A 5 means the work was accepted as delivered. Middle scores mean it needed revision. A 1 means it failed.
The bar is professional work. Every result is judged against what a paid professional would have delivered for the same brief.
Tasks stay private so they can't leak into training data and inflate future scores.
The goal isn't one more leaderboard. It's helping you pick the right model for your kind of work. A model that leads in reasoning can still fall behind in writing or design taste, and the radar charts show exactly where.
This is v0.1. It will keep evolving with harder, market-relevant tasks and with new models as they ship. A few popular models (Opus 5, GPT Luna) aren't included yet because I haven't run enough tasks on them to score them fairly.
Da7em Bench is fully independent. No sponsors, no vendor relationships. I've paid for every run out of pocket, thousands of dollars so far. That's the whole point: an honest, neutral look at what these models actually do on real work.
Full scores and the framework are in the images below.
7
6
78
6,048
I like Astra/Sol, but... The usage just drains too quick. The truth of it is, we pay $200/month for a subscription to use the best models. It is no longer worth it when you can't actually use the best models. 1-2 days of usage on Astra/Sol when I get similar usage on ? I'm not even a big fan of Anthropic. :\
0
0
1
27
Did Kimi change their 5-hour usage quota? I've done two minor changes with Kimi K3 and haven't maxed out my usage - amazing!
0
0
0
38
I'm so done with Anthropic... You can't use Fable because:
1). It's too expensive and eats up your 5 hour limit doing simple tasks.
2). Most of the time it just rolls back to Opus anyway.
For the first time, my Claude subscription is going unused until the expiration date and I won't be renewing.
1). It's too expensive and eats up your 5 hour limit doing simple tasks.
2). Most of the time it just rolls back to Opus anyway.
For the first time, my Claude subscription is going unused until the expiration date and I won't be renewing.
0
0
0
11