Your Claude bill called. It wants to talk.
|
This post was written with the help of AI. |
AI-assisted coding is great. Until you check the bill.
The Claude Code docs cover how to reduce costs. This post ranks cost reduction practices by impact and tells you which ones to do first. Most developers know a few. Fewer apply them consistently. The ranking weighs three things: how much a practice saves per session, how universally it applies, and how easy it is to start. The first seven work for everyone. The last three depend on your setup.
Each of the 10 practices below is worth 2 points: 1 for knowing it, 1 for applying it. Keep score as you read.
|
This post is focused on Claude. Some of it may apply to other leading AI models. Local models can also be a valid option to lower costs, but out of scope here. |
1. What’s the single biggest lever to reduce costs?
Model selection. First point: free. Second point: when did you last reach for Haiku?
Pick the right model for your task:
-
Haiku: simple and fast operations. Formatting, summarization, quick lookups. No heavy reasoning needed.
-
Sonnet: most coding tasks. The sweet spot between capability and cost.
-
Opus: complex reasoning, architectural decisions, tricky bugs. Reach for it when the task genuinely needs it.
-
Fable: top of the range, currently unavailable. Most capable, most expensive. Use it when nothing else will do, or when it’s not your money.
In interactive sessions, use /model to switch. In skills and agents, set it in frontmatter.
---
model: sonnet
---Dig deeper: Models pricing, Model configuration, Skill frontmatter, Sub-agent frontmatter
2. Your agent is thinking hard on every task. Is that a problem?
Yes. Claude is writing a dissertation about your variable renaming.
Thinking tokens count as output tokens. The priciest kind. A single request can burn tens of thousands of them. For a complex architectural decision, worth every token. For "add a missing semicolon," not so much.
In interactive sessions, use /effort low or /effort medium for straightforward tasks, and /effort high when it actually matters. In skills and agents, set it in frontmatter so nobody has to remember.
---
effort: low
---Dig deeper: Models pricing, Effort configuration, Skill frontmatter, Sub-agent frontmatter
3. What does 'improve the codebase' actually cost?
A lot. The fix is simpler than you think: be specific.
Claude starts by figuring out what your codebase even is. That means reading files until it has enough context to act. And every file it reads stays in context, getting re-sent with every follow-up message. You’re paying for that exploration whether you asked for it or not.
Specificity isn’t just for interactive sessions. In skills and agents, every word in the spawn prompt loads on top of everything the agent auto-loads, from turn one. Vague instructions are just as expensive there.
Free points. You’re welcome.
4. Your CLAUDE.md documents your entire architecture. Smart or not?
Impressive documentation. Expensive documentation.
Everything in CLAUDE.md loads at session start and gets re-sent on every single request, whether it’s relevant or not. That beautifully written section explaining the history of your repository pattern? Claude reads it before fixing a typo. Every. Single. Time.
-
Keep CLAUDE.md under 200 lines. Essentials only.
-
Move path-specific guidelines into rules. They load only when Claude works on matching files, not on every turn.
-
Move workflow instructions (PR reviews, deploy steps, migration guides) into skills. They load on-demand, not on every turn.
Dig deeper: Path-specific rules, Extend Claude with skills
5. You just finished a task. What should you do before starting the next one?
/clear. Seriously.
Every request re-sends the full conversation history as input tokens. Once a task is done, that history is dead weight: old file reads, debugging logs, that long exchange where Claude went down the wrong path. You’re carrying a suitcase full of things you don’t need anymore.
/clear drops it.
6. You’re mid-task and the context is getting heavy. What do you do?
/compact, but with instructions.
Just running /compact summarizes the conversation, but Claude decides what to keep.
You can do better: /compact Focus on modified files and unresolved errors tells Claude exactly what matters.
You can also make it permanent in CLAUDE.md so it applies every time Claude compacts automatically.
== Compact instructions
When compacting, focus on modified files and unresolved errors.Dig deeper: Compact with a focus
7. Your agent is running tests and reading logs. Where does all that output go?
In the main context, if you’re not careful.
Test runners, log processors, and doc fetchers can generate a lot of output. If that output lands in the main session, Claude re-sends it on every subsequent turn. Subagents run in their own isolated context window. The parent session gets a summary, not the raw output.
In skills and agents, structure workflows to delegate verbose operations to subagents. The main agent stays lean. In interactive sessions, Claude will usually make that call on its own, but you can always ask explicitly.
Dig deeper: Isolate high-volume operations
8. Scripts and hooks helping Claude process less
|
This one is situational: the savings depend on how much output you’re actually cutting. |
Pre-written scripts and hooks let you preprocess data before Claude sees it. Feeding Claude a raw test report when all you needed was the failures means Claude processes a lot more than necessary.
A 2026 study filtered low-value output from agent trajectories and found 40-60% fewer input tokens and 21-36% lower costs, with no impact on task success. Scripts and hooks are a simpler way to act on the same idea. The pattern scales well for agent-heavy workflows, but for small tasks the overhead isn’t worth it.
Dig deeper: AgentDiet, FSE 2026
9. Claude is about to rewrite half your codebase. Has it seen a plan first?
It should have.
|
This one is situational: it only pays off on complex or ambiguous tasks. |
Skip the plan and Claude might build the wrong thing entirely. You pay for the wrong implementation, then pay to fix it. Plan mode makes Claude explore and propose an approach before touching any code. You review it, adjust it, then let it implement.
In interactive sessions, press Shift+Tab before giving Claude a large or ambiguous task. In skills, bake it in: structure the definition to include an explicit planning step before execution.
Dig deeper: Plan mode
10. Did you actually need that 1M context window?
|
This one is situational. |
Surprise: it’s not free.
The 1M context window costs the same per token as the default. It just holds 5x more tokens, so sessions stay uncompacted much longer and every turn carries a heavier payload. Stick with 200K unless you actually need it.
Dig deeper: Extended context
How do you know if it’s working?
You can’t improve what you can’t measure.
Run /usage in Claude Code to see token usage and cost for the current session.
Watch those numbers across sessions as you apply these practices.
The difference is visible.
Running the API or tracking team usage? The Claude Console breaks it down by model, date, and API key.
Dig deeper: Using the /usage command, Claude Console
How many points did you get?
Scored 15 or above? Genuinely impressive.
Below that? You’re one or two practices away from making a real difference. Start with model selection, and go from there.
Thanks for reading!
Leave a comment