Think things through, work out the numbers, or start with a question. Choose your model. Keep control.
Prebuilt use cases
Choose an example to fill the composer, edit it, then Send. Examples use synthetic data; review every model answer.
Local workspace
One document. Manageable pieces.
Make room to understand.
Split UTF-8 .txt or .md files up to 10 MiB on this device. Manual sends upload one selected chunk. Optional Temporal processing uploads the whole document after confirmation. Public or synthetic data only.
No document loaded. Local preview does not save documents or chunks.
Each manual send starts a fresh context. Results appear in Conversation and count toward token usage. Manual sends do not batch chunks or generate a combined summary.
Process with Temporal.
Automatically process every chunk, one at a time, with the verified local model selected in Conversation. Each chunk uses the processing instruction above and a fresh context. Results remain separate; no combined summary is generated.
Checking Temporal availability...
Understand your compute
Every token, accounted for.
Provider-reported usage across model rounds, saved on this machine. No prompts or answers are stored in the usage history.
Reported input tokens
0
Reported output tokens
0
Reported total tokens
0
By model
Model
Calls
Input
Output
Total
Unreported
Recent model calls
Time
Model
Purpose
Input
Output
Total
Status
Counts include tool-call rounds and, when selected, readiness probes. Calls with missing counts are marked unreported, not estimated as zero. Pending calls may have been interrupted by a restart. This is not a billing statement.