# Cost What a feature actually costs to build with AI agents, and how to keep spend visible instead of finding out at the invoice. Source: https://codeherder.com/blog/topics/cost/ Topic Every agent run has a price. See it per feature, not just at the end of the month. Cost, here, means the token spend and compute time behind a shipped change. We track it down to the feature and the task, not just the account total. CodeHerder logs every session's spend so a $4 bug fix and an $80 feature are both visible before they surprise you. This matters because agent spend does not scale like a headcount. A model upgrade, a bad prompt, or a task that loops turns a five-minute job into an expensive one. A monthly bill will not tell you which task did it. Line-item visibility does. Read this hub for real cost breakdowns from tasks we ran ourselves. It also covers the levers that move spend the most, and how to set a budget an agent respects. [Request access →](https://codeherder.com/get-started/#request-access) September 29, 2026 · 6 min read ## [We filed the same vague brief 84 times](https://codeherder.com/blog/the-same-vague-brief/) One short, vague brief, run 84 times by agents. Median cost about $18, median time 90 minutes. Here is what shipped and what it cost. For everyone - [workflow](https://codeherder.com/blog/topics/workflow/) - [cost](https://codeherder.com/blog/topics/cost/) - [product](https://codeherder.com/blog/topics/product/) August 16, 2026 · 8 min read ## [Optimise the file every agent reads before doing anything](https://codeherder.com/blog/the-file-every-agent-reads/) Keeping CLAUDE.md small is best practice, but ours hit 231KB across 37 correct commits. Splitting it cut cost per turn 28%, turns per task by a third. For everyone - [workflow](https://codeherder.com/blog/topics/workflow/) - [cost](https://codeherder.com/blog/topics/cost/) - [observability](https://codeherder.com/blog/topics/observability/) August 5, 2026 · 8 min read ## [The most expensive line in your agent config is the model name](https://codeherder.com/blog/reasoning-model-is-the-cost-lever/) Agent cost doesn't track codebase size. Swapping our judgment stage's reasoning model moved a story's cost 2.4x, and its lead time nearly 3x. For everyone - [cost](https://codeherder.com/blog/topics/cost/) - [orchestration](https://codeherder.com/blog/topics/orchestration/) - [observability](https://codeherder.com/blog/topics/observability/) August 1, 2026 · 10 min read ## [Give your agents their own machine: what CPU contention costs a parallel herd](https://codeherder.com/blog/dedicated-hardware-for-concurrent-agents/) Parallel Claude Code sessions need dedicated hardware. Load-matched data puts one machine 40% ahead, and a spot m9g.xlarge holds ten sessions. For everyone - [orchestration](https://codeherder.com/blog/topics/orchestration/) - [cost](https://codeherder.com/blog/topics/cost/) July 26, 2026 · 14 min read ## [Anatomy of an $8 feature: the itemised receipt for code no human wrote](https://codeherder.com/blog/anatomy-of-an-8-dollar-feature/) 944 user stories shipped in a week, with no human writing the code. Here's what the median one cost, and what each quality gate caught. For everyone - [cost](https://codeherder.com/blog/topics/cost/) - [workflow](https://codeherder.com/blog/topics/workflow/) - [quality](https://codeherder.com/blog/topics/quality/) - [observability](https://codeherder.com/blog/topics/observability/) July 25, 2026 · 14 min read ## [10 things we learned running 100 billion tokens through a herd of AI agents](https://codeherder.com/blog/100-billion-tokens/) Two months of CodeHerder data: 1.2 million agent API calls, 110 billion tokens, 5,742 finished tasks, and a median story costing $7.54, done in 50 minutes. For everyone - [cost](https://codeherder.com/blog/topics/cost/) - [observability](https://codeherder.com/blog/topics/observability/) - [orchestration](https://codeherder.com/blog/topics/orchestration/) June 1, 2026 · 3 min read ## [Cost-aware AI engineering: keeping agent spend visible and bounded](https://codeherder.com/blog/cost-aware-ai-engineering/) Token spend is easy to lose track of with more than one agent running. Here's how to keep it visible per task, agent, and model. For everyone - [cost](https://codeherder.com/blog/topics/cost/) - [observability](https://codeherder.com/blog/topics/observability/) - [quality](https://codeherder.com/blog/topics/quality/) [← All topics](https://codeherder.com/blog/topics/) [All insights](https://codeherder.com/blog/)