Live recording analysis

Debug info

SessionReady

Events0

Actions0

Evidence0 KB

API spend$0.000000

LunaWaiting

Compared with VLM—

Waiting for events

Live updates repeatedly analyze the recording as it grows. After-recording processing creates one final summary. While recording, the final-summary estimates cover the recording so far.

Scenario	Luna / LLM API	VLM API	Video render	VLM total
Live updates Updates so far	$0.000000Reported spend	—	—	—Estimate
After recording One summary per recording	—Rough estimate	—	—	—Estimate
1,000 similar recordings One summary each	—	—	—	—

Live spend includes all update attempts; local compute excluded.

The 1,000-recording row multiplies this recording's one-time estimate by 1,000. It is not a forecast for recordings of different lengths. Post-recording processing uses standard scenario rates, not a provider Batch API discount. VLM estimates include fresh rendering; LLM extraction and all other infrastructure costs are excluded.

Scenario estimates only. No video is rendered and no VLM requests are sent.

Assumptions & breakdown

Rates are checked-in scenario assumptions, not a live price lookup. One uncached call and one fresh render per update; no caching or render reuse. The model label does not change when rates are edited.

Waiting for recording…

Waiting

Open a recording browser and interact with the page,
or load an existing events.json. The replay will appear here.

Revision 0

The DOM persists across batches. The latest action outcome remains provisional.

Waiting for recorded events…