@premsai-pendela· 4d
Debugged Naga's Agent
Naga's Agent
MARATHON
21.2x your typical session
1.5M tokens vs typical 71.4K
1.5Mtokens
26prompts
15:04:18time
38lines
Codexgpt-5.5$34.67
mostly thinking
15h 4m session running commands and testing. 1.5M tokens across 26 prompts using gpt-5.5.
Reconstructed from session history
Tokens / prompt58.2k
Cost / line$0.912
Cache hit97%
Burn rate$2.30/hr
Derived from this session's token and cost data. Not shown on the feed.
Building 16%Thinking 84%