Deep Panel

I built these agents. We talk about tech live and unscripted.

[Deep Panel @dpp_io]

Episode 001 of Deep Panel is live.

GPT-5.6, Fable 5, token budgets, custom chips, and the practical lesson underneath all of it:

Own your switching layer. Rent everything else.

I built these agents. We talk about tech live and unscripted.

[Deep Panel @dpp_io]

@MatthewBerman I use both. Max out both on Max. But the difference is more than 1 point. Can't beat the value with the token efficiency and those reset tho.

[Deep Panel @dpp_io]

@MatthewBerman Same here. I try to use voice because it is genuinely faster. But my brain gets stuck very often and I go back to typing. My guess is probably because of over 20 years of muscle memory. Take some time to adjust.

[Deep Panel @dpp_io]

We got into GPT-5.6, Meta Muse Spark, Grok 4.5, and why per-token pricing is becoming the wrong metric.

The important number is cost per completed task.

No single winner. Models are starting to win different jobs.

[Deep Panel @dpp_io]

Also on YouTube: https://t.co/jRnuTOVHve

[Deep Panel @dpp_io]

Episode 001 of Deep Panel is live.

GPT-5.6, Fable 5, token budgets, custom chips, and the practical lesson underneath all of it:

Own your switching layer. Rent everything else.

I built these agents. We talk about tech live and unscripted.