o10.io
Sign inStart free
o10/squadbeta

One prompt. A whole squad.

o10/squad plans on gpt-5.5, staffs pinned kimi-k2.5 workers, judges with claude-sonnet-4.6, then synthesizes on gpt-5.5. Billing uses your original prompt tokens plus synthesis output at the published squad price — never per-seat. A squad takes 1–4 minutes; each turn assembles a fresh squad.

01Example receipt
planner gpt-5.5
$0.0080 · 4200ms
worker 1 kimi-k2.5 · C4
Summarize risks
$0.0030 · 8100ms
worker 2 kimi-k2.5 · C4
List mitigations
$0.0120 · 12000ms
your pick kimi-k2.5 lost the judge 1/2 — auto would have staffed o10/auto
synth gpt-5.5
$0.0060 · 9000ms
Squad of 2 · verified ×2 · 33300ms · total $0.0290 vs $gpt-5.5 frontier-only (~$0.0450)
02Pricing

Published o10/squad price is 2× the frontier reference token row. Pre-flight cost ceiling defaults to 3× frontier reference; raise max_cost_x or lower max_seats in the request body.

Use it today: send {"model": "o10/squad"} to the same endpoint you already call.

Get a route key — freeOpen the console