It looks pretty consistent? At least within reason for a stochastic model. Or am I missing something?

What you may be missing is that they probably track the API performance.

Subscription plans may be subject to other regime, e.g. lowering the thinking budget when the API is under heavy load, etc.

Low is good if you’re working in a tight loop. But more risky for more agentic stuff you want to let cook for 30 minutes or more.

I always work on a tight loop, I don't think it makes sense to let agents run for hours.

Regardless of the model, running it for hours means that the model will takes decisions and assumptions alone instead of you.

Yup that’s quite literally what ‘agency’ is and the whole point of agentic workflows. Personally I’ve had them running for days with good results and as you can see OpenAI had them running for months, and yes indeed the things made some very questionable decisions and assumptions… but they unquestionably did a lot of stuff correctly, for some definitions of ‘technically correct’.

Personally I don't believe in agentic workflow. I don't think that's a coincidence that both OpenAI and Anthropic chose math problems to test their long agentic workflows, they are well defined, with a clear finish line and with a 100% clear progress path, most of real life tech projects aren't like that.

Due to this release note, I used it once to rewrite a doc, and was disappointed.

Ya it’s annoying to have to manage this.

But effort is basically how much extra internal scratchpad to use and how much extra questions to ask and answer before producing a result, Exploring more hypotheses, validating consistencies, calling more tools.

If you’re happy with your token spend on Astra then keep doing what you’re doing. but if you feel the need to conserve tokens, then you can do that by switching to smaller models like Luna when the task is straight forward.

I feel the need to self sensor my writing when phrases sound like AI. Pretty annoying.

It's now a CAPTCHA to tell humans apart from bots based on how you write.

You could try throwing in the odd homonym in place of an important word so that people know it must have been written by a human.

I leave typos in all my writing so people know it’s not ai

It misses the forest for the trees. It feels the need to highlight details not understanding what details are most relevant to a human reader and how to survey the larger problems in a cohesive way that emphasizes the right parts without cliche and undue emphasis.

Would Jev be good for this?

It can play doom in realtime so maybe it can play this.

Did you see the Jev model demo where played wiki racing against GPT, Claude and gpt hallucinated links in Wikipedia pages?

Even worse is having to telling people to do stuff who then turn around and use AI to do a poor job without the human rationalizing the problem or result.

Oh wow I remember having the same feeling there. I had forgotten about that, thanks for the memory.