iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Not necessarily—but the headline needs two important qualifications. Uladimir Klyshevich says he used 40 billion tokens over three months to build the open-source Dart agent harness >_Fa, while spending about $1,050 on subscriptions. The roughly $17,000–$20,000 figure is his estimate of what the usage might have cost at API list prices, not the cash bill he reports paying. Both figures and the project claims come from his own account, not an independent audit. Klyshevich’s account
What the 40 billion tokens and $20,000 figures mean
Klyshevich’s headline claim is “40,000,000,000 tokens. 3 months of work. Zero lines of code written by a human.” He says he built >_Fa, an open-source agent harness written in Dart, using a mix of Kimi and GLM models routed to different tasks. The account is self-reported; it does not include an independent usage audit or code review. Klyshevich’s account
| Figure | What Klyshevich says it represents | Qualification |
|---|---|---|
| 40 billion tokens | Approximately 24 billion cached-read tokens, 14 billion fresh-input tokens, and 2 billion output tokens over three months | Author-reported project usage; not independently audited |
| $17,000–$20,000 | Estimated API list-price equivalent for the stated token mix and rates | Author’s estimate, not a bill or verified cost calculation |
| About $1,050 | Reported subscription payments for two Kimi subscriptions and two GLM subscriptions | Author-reported cash expenditure; receipts are not supplied |
He says subscriptions, caching, and routing work to lower-cost models explain why the reported cash outlay is much lower than the API-price estimate. His post does not provide usage exports or a reproducible calculation, so the gap is plausible as his explanation but cannot be checked from the figures alone. Klyshevich’s account
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Scan for outdated or missing drivers - takes under a minute3Clear out junk files and repair common Windows errorsWhat he says he built
Klyshevich describes >_Fa as a Dart agent core with streaming providers, tools, sessions, compaction, and memory. He says it has command-line, desktop, mobile, and web interfaces, and describes browser-extension and CI use, on-device or browser-side execution, sandboxed scripting tools, git-backed memory, agent-to-agent messaging, session replay, runtime widgets, and declarative sandbox profiles. These are descriptions in his post, not independently tested feature findings. Klyshevich’s account
#1 Best Overall
He also says a separate factory project, dmtools-dart, builds and ships >_Fa releases, and that this release process has run without a human in the loop. That is an automation claim, not evidence that the resulting software is secure, reliable, or maintenance-free. His post reports an approximately 7 MB CLI binary; any comparison of installation sizes is scoped there to macOS arm64 artifacts measured in September 2026, not a universal or independently reproduced benchmark. Klyshevich’s account
Does the project make the spending rational?
There is no single answer in the available evidence. “Am I crazy?” bundles together three different questions: what Klyshevich paid, what the same token usage might cost through API list pricing, and whether the resulting software is valuable enough to justify the time and expense. The first two have author-reported numbers; the third is not measured.
- Cash outlay: He reports about $1,050 in subscriptions, without supplying receipts.
- API-price comparison: He estimates $17,000–$20,000 for the stated token mix at API list rates; this is a counterfactual pricing comparison, not proof of savings against a verified bill.
- Project value: He describes a broad feature set and automated build-and-release workflows, but the post does not independently establish software quality, user value, or return on investment.
The project could make sense as an experiment for someone who values the software, can use subscription access within its terms, and can assess and maintain agent-generated code. The account does not quantify human supervision, failed work, maintenance, opportunity cost, or ongoing operating costs. It therefore does not show that other teams could reproduce the result or save money by following the same approach. Klyshevich’s account
Recommended Free Tools
How to assess the claim without treating token volume as a verdict
Klyshevich describes the reaction to the figure this way: “When people hear the number, they hear a fuckup. Forty billion tokens sounds like setting money on fire.” That is his characterization, not survey evidence about what developers generally think. Token count alone cannot establish waste: it says little about how much useful work resulted, how much was repeated, or how much human effort was still needed.
If you are evaluating this project—or considering a similar AI-heavy build—separate the questions that the headline compresses:
- Cost: distinguish cash actually paid from a list-price estimate, and ask what access model and usage assumptions produce each number.
- Output: inspect the code, tests, security boundaries, supported platforms, and maintenance needs rather than treating a feature list as proof of quality.
- Repeatability: check whether the model access, routing, caching, and workflow can be used under the relevant subscriptions’ terms and reproduced by your team.
- Comparison: when comparing harnesses, distinguish installation size from runtime dependencies, platform coverage, capabilities, security, provider support, and maintenance. A single platform-specific artifact-size comparison cannot settle which tool is better.
The post supplies an intriguing first-person case study, not a neutral benchmark. It does not establish audited spending, independently verified features, general return on investment, or repeatability for other teams. Klyshevich’s account
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →
