Compare the top 3 LLM APIs and recommend one
Notes rise with tokens spent; the figure beside each note is its token count; one bar is one parallel step.
- Elapsed
- 0.00 s
- Agents
- 0/5
- Tokens
- 0
- Cost
- $0.0000
- 1
- 2
- 3
- 4
- 5
The ledger
Each finished run, newest first. Open one for its full trace: every step, its bar, when it ran, its tokens and its cost.
- Runs
- 0
- Tokens
- 0
- Cost
- $0.0000
- Time played
- 0.00 s
No runs have finished yet. The first one is playing in the score above; when the critic signs, it is entered here.
Personnel
Every agent that played this session, by tokens spent.
The personnel list fills in as runs finish.
Instruments
The four tools, with a tally of every call this session.
- web_search0 calls · 0 tok
Finds and reads sources on the open web.
Not called yet this session
- structured_output0 calls · 0 tok
Normalizes what was found into tables and typed fields.
Not called yet this session
- run_code0 calls · 0 tok
Executes analysis, tests and scans in a sandbox.
Not called yet this session
- generate0 calls · 0 tok
Drafts prose, plans and code.
Not called yet this session
The budget
The current run against its token cap, and every finished run against the cap it ran under. Costs use a sample price of $0.003 per 1K tokens.
This run
Scored for 1,940 tokens in 4 bars, 560 under the cap, so the guard lets every bar play.
This session
Finished runs are drawn here, one bar each, against the cap they ran under.
