Divide the endpoints and the number gets bigger. 292,000 tokens per solved task for OpenClaw against roughly 3,500 for Aider in architect mode is a factor of about 83 [6][1], not the 70-fold in the writeup's own headline [23]. Round numbers travel better than division does. Either figure sits far outside anything a buyer wins by negotiating price per million tokens.
The floor recurs because of how serving works. Every inference request needs its context, and the relevant history is either supplied again or reconstructed by the serving system, so the provider processes large overlapping blocks of text on each turn, including the harness's system prompt and its tool descriptions [2]. The startup tax is that baggage measured once: the system prompt, the tool descriptions and the environment setup, around 700 tokens for Aider in architect mode against around 26,000 for OpenClaw [7]. The June author's regression multiplies it by turn count and predicts tokens per solved task with an R-squared of 0.99 across both models [9].
You can run that regression backwards. Divide 292,000 by 26,000 and you get about eleven [4]. So on the tasks OpenClaw solved, roughly eleven resends of the same scaffolding account for most of the token spend, and the actual task content is what is left over. The fifteen-turn illustration in the post, 390,000 input tokens of scaffolding [8], is above the measured per-task total, which is a useful reminder that it was an illustration and not a measurement.
Dollars did not spread as far as tokens. Composio's eight-harness comparison put cost per successful task between $0.028 for Pi Agent and $0.195 for Claude Code, a range of about 7x [16][3]. Different harnesses, and different work: thirty workflows across Airtable, Gmail, Google Calendar, Google Sheets, GitHub, Slack and PostHog, each under a 900-second ceiling [13]. DeepAgents matched Claude Code's pass rate exactly at a quarter of the cost per success, so around five cents [17][5]. Composio graded with a programmatic verifier rather than an LLM judge, on isolated fixtures seeded with decoys and near-identical keys [14], and it disclosed that Pi ran a different reasoning setting across two model providers and that Prime Agent produced only 24 gradable runs out of 30 [18]. That is careful benchmarking, and it is also why the result cannot be read as single-variable.
For the token ordering to transfer to your invoice, your turn distribution has to resemble theirs. The June suite was twelve Python tasks with one pass per harness, task and model combination, so it carries no variance estimates while agent runs are stochastic [3][10]. Artificial Analysis is the better instrument for ranking live pairings, at 326 tasks with pass rates averaged over three attempts, plus a controlled swap that holds Claude Opus 4.7 fixed across Claude Code, Cursor CLI and Opencode [19][21][24]. What transfers without any benchmark is the method: measure your harness's prompt floor, then your median turns to done [11].
