Economics of agentic coding
Know what AI tokens buy you.
Token economics is the market and operating system around AI tokens: how compute is priced into tokens, how organizations allocate them, how agents consume them, and how that spend is measured against outcomes.
Vendor billing is designed around usage totals. It does not explain what came from context, cache traffic, retries, tool output, subagents, or model choice. Much of the bill can come from scaffolding around the model, not the model's reasoning. Even cutting tokens can raise cost when it breaks the cached prefix. Teams know what they owe, but not what they bought.
Quesma turns coding-agent sessions into the itemized record: what consumed tokens, what the work accomplished, and which costs teams can change. It surfaces specific optimization opportunities and shows whether each change reduced cost without reducing useful work.
Same tree, fixed
Every failure mode maps to one specific practice. Switch the whole system, not isolated leaves.
Select a branch to inspect its paired failures and practices.
Evidence ledger
input-token spread, same model, leaner harness scored higher
Terminal-Bench, 2026 ↗4.7xClaude Code ~32,800 vs OpenCode ~6,900 scaffolding tokens
Systima, 2026 ↗≈80%of a real provider bill came from cache traffic
arXiv:2607.12161, July 2026 ↗−38% / +6.8%tool-output tokens removed, cost increased after the cached prefix broke
arXiv:2607.12161, July 2026 ↗≈0.1xbase input price for cache reads on both major providers
Anthropic and OpenAI pricing, 2026-08 ↗≈2xbill after a Codex CLI compaction retune
Codex CLI v0.118 regression, 2026 ↗$1,434/dayconsumed by a documented retry storm
arXiv:2604.16646, 2026 ↗$1,500/moper-tool cap at Uber after about four months of budget burn
TechCrunch, 2026 ↗+7.6%cost at low effort in an A/B test of a plugin claiming 60–90% cuts
JetBrains, July 2026 ↗Maintained in the open
The efficient tree is a curated, evidence-graded practice set, maintained in the open.
Field notes
HN How I burned a full Claude limit in 30 minutes and built my own deep research pipeline instead: 3 subscriptions, shared memory, a clear role for each model. You can build the same from what you already pay for.
We ported the puzzle game Baba Is You to the Harbor framework, and benchmarked current models, including Claude, GPT, Gemini, GLM, and DeepSeek. A human Twitch streamer is 4x faster than Claude Fable 5.
Featured Tokenflation: simple tasks consuming ever more context, reasoning, and tool calls without more useful output. We benchmarked 14 models; one answered “Hi” with 33 tool calls and an unsolicited commit.
