Unverified50% confidenceBenchmarkExact time
ReAct consumed an average of 4,663 tokens per question, Plan-execute 10,535 tokens, Supervisor 7,494 tokens, and Draft-verify 4,905 tokens.
1
Sources
50%
Confidence
Long-term
Relevance
10/17/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedEach retry resends the complete prompt input tokens and is billed in full; a request with 4,000 input tokens retried 3 times will consume 16,000 tokens.74% similarUnverifiedThe prefill throughput of this setup increases with prompt length, climbing from approximately 290 tokens/s for short prompts to approximately 1,300 tokens/s, with an average prefill throughput of approximately 900 tokens/s.73% similarUnverifiedToken generation is billed by consumption; if an application requests a 4,000-token completion and the user stops after the 100th token, failure to propagate the cancellation signal means paying for the remaining 3,900 unrendered tokens.72% similarUnverifiedThe tester noted: Cursor did not report cache-write tokens, so its true token consumption is very likely higher than 92,000.68% similarUnverifiedAccording to an analysis based on real API traffic, out of about 411,000 requests over 7 days, 99.3% of processed tokens were input tokens; the median request had 3,782 input tokens and returned only 147 tokens.66% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1053856API
curl https://kongchang.com/api/v1/knowledge/claims/1053856MCP
get_claim(id=1053856)