Which Claude should you actually pay for?
Do not pick one Claude for everything: plan with the most capable model, execute with a strong cheaper one, and let it hand the small pieces to sub-agents.
What the reel showed
I gave Claude Opus 5, Sonnet 5, Haiku 4.5 and Fable 5.1 the same brief: an airport seen from overhead that runs itself. Planes land, taxi, park at a gate and take off again, with a departures board that updates, all in one web page sized for a 1000 x 1000 window. One attempt each.
Haiku finished in under a minute and a half for 12 cents, and its airport does not fit the frame: a runway runs off the edge and the departures board is cut off. Sonnet built a clean, working plan view for $1.66. Opus built a dark radar screen for $3.00.
Fable took almost 19 minutes and $7.00 for the most detailed airfield: grass, a coastline, jet bridges, a hangar, a fire station, and separate arrivals and departures boards.
| Entrant | Tool | Time | Cost | Verdict |
|---|---|---|---|---|
| Claude Opus 5 | Claude Code | 11m57 | $3.00 | Dark radar screen, works |
| Claude Sonnet 5 | Claude Code | 13m43 | $1.66 | Clean plan view, works |
| Claude Haiku 4.5 | Claude Code | 1m28 | $0.12 | Does not fit the 1000 x 1000 frame |
| Claude Fable 5.1 | Claude Code | 18m50 | $7.00 | Most detailed airfield |
All four ran in Claude Code with the same brief, one attempt each, and each cost is read off that run's own bill.
What this means for your business
The reel is the evidence. The same airport cost 12 cents on Haiku and $7.00 on Fable, a 58 times spread in the same tool. Haiku's did not fit the frame, so the cheapest model was not enough for the whole job. Sonnet and Opus both delivered a working airport for well under half of Fable's bill, so the most expensive model was not needed for every part of it either.
The answer is not to find the one right model. It is to give each model the part of the job it is best at. Planning is where a mistake costs the most, because every later step builds on it, so that is where the most capable model earns its price. Carrying out a clear plan is well within a strong everyday model. Small, well-defined pieces, like drafting one section or checking one file, can go to cheaper models working in parallel.
That is agent orchestration, and it is how I run almost every job now. The planner decides what needs doing, in what order, and how to check each part. The executor follows the plan and hands the bounded pieces to sub-agents. You pay top-model prices only for the thinking that needs it, and get the best result per token.
It works for business tasks, not only code: a website refresh, a month of social posts, a pricing page, a client report. Anything with more than a few steps comes out better with a plan written before the work starts.
What I’d use
My actual setup: I plan with Claude Fable 5.1. Once the plan is done, I switch to Opus to execute, and Opus spins up its own sub-agents for the independent pieces. That is the main takeaway from all of these tests: the best model for each job, used in the right order.
I run this on subscriptions, not a bill per call. Claude Code on my Claude plan and Codex CLI on my ChatGPT plan are what I use every day. I only go to OpenRouter when I want to compare several models side by side, the way this reel did.
If Haiku or Sonnet is all you use today, keep them for the sub-agent work. Bounded pieces with a clear check are where cheap and fast pays off.
Try it yourself
Run one real business task the way I do: plan on your most capable model, execute on a strong everyday model, and let it hand the bounded pieces to sub-agents. All of it runs in Claude Code on a Claude subscription.
Get Claude Code on your Claude plan
Claude Code is included with Claude Pro and Max. Install it and sign in with your Claude account, and your runs draw on your plan's usage limits instead of a bill per call. Fable is included on Max; on Pro it runs on paid usage credits, so on Pro you can plan with Opus instead.
Start the session on your planning model
Make an empty folder for this task and start Claude Code in it. Type /model and pick your most capable model from the list, for example Claude Fable 5.1.
Switching models in Claude Code
/model
Paste the planning prompt
Fill in the brackets with a real task from your business. The prompt tells the model to plan only, and to mark which steps can run in parallel and which are simple enough for a cheaper model.
I want to plan a piece of work before anyone starts on it. Do not write code, files or final content yet. Only plan. The task: [describe the task in two or three lines, e.g. "Refresh our services page and add a pricing section."] Who it is for: [your customers, your team, a client] The one thing that must work: [e.g. "prices match our current price list"] What done looks like: [how we will know it worked] Constraints: [budget, tone, tools we use, anything that must not change] Give me a written plan with: 1. Scope: what is in, what is out, and any questions you need me to answer first. 2. Steps, in order, each small enough to finish and check on its own. 3. Which steps can run in parallel because they do not depend on each other. 4. Ownership: for each step, which files or sections it creates or changes, so no two parallel steps touch the same thing. 5. A check for each step: how we will know it is done and correct. 6. For each step, whether it needs careful judgment or is simple enough for a cheaper, faster model. If anything is unclear, ask me first. Then write the plan and stop.
Read the plan and fix it now
Answer its questions and correct anything that is wrong or missing. This is the cheapest moment to catch a mistake, before any work is built on it.
Switch to your execution model
Type /model again and pick a strong everyday model, for example Claude Opus. The plan stays in the conversation.
/model
Paste the execution prompt
This tells Opus to follow the plan, give the independent, bounded steps to sub-agents on a cheaper model, and keep the judgment calls and the final review for itself. Claude Code can start sub-agents and choose a model for each one.
Now carry out the plan above. You are the lead on this job. - Follow the plan step by step. If you find a problem with the plan, stop and tell me before you change course. - Hand the steps the plan marks as independent to sub-agents, running in parallel where they do not overlap. Where a step is bounded and easy to check, run that sub-agent on a cheaper, faster model. - Give each sub-agent a short brief: exactly what to do, which files or sections it owns, what not to touch, and the check that proves it is done. - Keep the steps that need judgment, putting the pieces together, and the final review for yourself. - When a sub-agent finishes, check its work against its check before you build on it. When everything is done, report: 1. what each sub-agent did, on which model, and whether its check passed 2. what you did yourself 3. anything left open, or that I should look at before I use the result
Check the result against the plan
Opus reports what each sub-agent did and whether its check passed. Go through the checks from the plan yourself before you use the result.
Want to compare models on one prompt instead?
To send one prompt to Haiku, Sonnet, Opus and Fable at once and see what each costs, the way the reel did, use OpenRouter's chat. One account and key for every model, pay per use.
Links
- Claude CodeAnthropic's coding agentclaude.com
- Claude plansClaude Code is included with Pro and Maxclaude.com
- Claude Fable on your planIncluded on Max, usage credits on Prosupport.claude.com
- Switching models in Claude CodeThe /model commandcode.claude.com
- Sub-agents in Claude CodeHow Claude Code hands work to sub-agentscode.claude.com
- Codex CLIOpenAI's coding agent, signs in with a ChatGPT plangithub.com
- OpenRouterOne account, every model, pay per useopenrouter.ai
- OpenRouter chatSend one prompt to several models side by sideopenrouter.ai
- Claude Haiku 4.5Fastest and cheapestopenrouter.ai
- Claude Sonnet 5The everyday defaultopenrouter.ai
- Claude Opus 5Stronger reasoningopenrouter.ai
- Claude Fable 5.1Top tieropenrouter.ai
Want this planning and execution setup running in your business?
I set it up around the work you actually do: which model plans, which one executes, which pieces go to sub-agents, and the prompts your team runs.