Made more improvements to the usage view in T3 Code. Much easier to see where your "spend" is and how different models behave based on your own data btw this measures all usage of Claude Code and Codex on your machines, not just usage within T3 Code :)

Can you use the data to figure out the averages for different subs and their value in like tokens per model and cost?

the cache savings line is oddly satisfying

hey @bil0090 quick question I dl’ed T3Codes yesterday (v0.44) but I don’t have the cool sidebar visual like Theo. How do you enable it ?

I live it :) Would be awesome to have a small always visible usage indicator in the sidebar! 👀

omg now i can like see my progress toward getting you banned in like a single dashboard isn't that like so cool

Tracking usage so you can feel guilty

At this point, UI/UX-wise, the only real rival T3 Code has is Codex... Codex feels incredibly polished and uniform. I bet they have crazy designers

i just updated 😭

that daily cost chart spiking like a heartbeat monitor is the real review

This is quite impressive

next feature request: a category for tokens spent undoing what the previous tokens did

does it show cost per session or just per model

But it doesn’t show if the model is nerfed like nerfbench? I would like a useless feature please. Thanks!

Thanks! Needed this

Spend per model on your own data beats any benchmark. Can it split by task type too? I'd like to know which model is cheapest for the boring edits vs. the planning.

tysm for this honestly. By far the easiest way to see how money is being spent.

Is my usage going down slower after switching to nightly or am I going crazy?

wait it counts claude code + codex outside t3 too? that's the spend view that actually matters

Started seeing this btw after updating to nightly yesterday. Installed providers are Cursor and Antigravity rn.

I want to use my reset within T3 Code 🥹

Usage dashboards are just postmortems with nicer charts and fewer incidents

measuring usage from the machine instead of one provider dashboard is the right call, seeing claude code and codex side by side is what i wanted

Add all harnesses (hermes too) and all accounts and wrappers man I might have burned hundreds of billions of tokens this year 😭

Does it show how close you are to the plan limit as well, or only the spend?

Great job

the “quick experiment” now comes with an itemized receipt

needed this before last month's claude invoice hit 😂😂😂😂😂

Finally a fuel gauge. On site nobody starts the engine without checking one. Took AI tools a while to catch up.

Thread specific usage?

Good tools need clear costs. It is nice to see how models behave with real data. Simple tracking helps build better habits. Smart design wins here.

measuring across tools is smart. what's the most surprising pattern you've seen in how people switch between Claude Code and Codex?

looks good

Use the usage view to compare models on your own recurring tasks, then route routine work to the cheapest model that clears your quality bar. Track all machine usage so the comparison isn't biased by one editor.