Claude Fable 5.1 in Claude Code: tokens per question
This newer Fable release remains in Claude's highest-priced tier for demanding coding investigations.
The figures
- Bare Claude Code
- 49,700
- With capsul
- 23,400
- less input sent
- x2.12
- answer checks passed out of five, bare then with capsul
- 3 / 4
- tool steps per question, bare then with capsul
- 5 / 2.4
- output tokens per question, bare then with capsul
- 1,570 / 391
- API-equivalent cost per question, bare then with capsul
- $0.636 / $0.279
cache-weighted input tokens per question
cache-weighted input tokens per question
90% interval: x1.72 to x2.65
Its fresh-input and output list prices match the earlier Fable row, while cached reads cost less. A long agent session may mix those categories, so the traffic split matters more than a single price label. This tier is most relevant when the task requires substantial reasoning and a cheaper model has not resolved it.
The measured row is partial: the subscription limit ended the run before all planned cells were collected. Its interval is especially easy to overread because the remaining sample is thin. Use the question-level evidence and answer checks, and do not generalise this row to an unmeasured workflow.
Question by question
Five one-line questions about a real TypeScript codebase, each asked 3 times per arm. Weighted input per question, mean of the repetitions.
| Question | Bare | capsul | Less input | Correct runs, bare / capsul |
|---|---|---|---|---|
| Q1 | 63,500 | 25,100 | x2.53 | 0 / 0 |
| Q2 | 35,900 | 20,000 | x1.79 | 2 / 2 |
| Q3 | 43,900 | 19,200 | x2.29 | 1 / 1 |
| Q4 | 60,700 | 26,800 | x2.26 | 0 / 1 |
| Q5 | 44,500 | 26,100 | x1.70 | 1 / 1 |
List prices
| Model | Input | Output | Cache read | Cache write 5 min | Cache write 1 h | Context |
|---|---|---|---|---|---|---|
| Claude Haiku 4.5 | 1.00 | 5.00 | 0.10 | 1.25 | 2.00 | 200,000 |
| Claude Sonnet 5 | 2.00 | 10.00 | 0.20 | 2.50 | 4.00 | 1,000,000 |
| Claude Sonnet 4.6 | 3.00 | 15.00 | 0.30 | 3.75 | 6.00 | 1,000,000 |
| Claude Opus 4.6 | 5.00 | 25.00 | 0.50 | 6.25 | 10.00 | 1,000,000 |
| Claude Opus 4.7 | 5.00 | 25.00 | 0.50 | 6.25 | 10.00 | 1,000,000 |
| Claude Opus 4.8 | 5.00 | 25.00 | 0.50 | 6.25 | 10.00 | 1,000,000 |
| Claude Opus 5 | 5.00 | 25.00 | 0.50 | 6.25 | 10.00 | 1,000,000 |
| Claude Opus 5.5 | 4.00 | 20.00 | 0.20 | 5.00 | 8.00 | 1,000,000 |
| Claude Fable 5 | 10.00 | 50.00 | 1.00 | 12.50 | 20.00 | 1,000,000 |
| Claude Fable 5.1 | 10.00 | 50.00 | 0.25 | 12.50 | 20.00 | 1,000,000 |
Anthropic list prices, checked September 23, 2026.
How this was measured
Two arms on the same model: Claude Code as anyone runs it, and the same question through capsul. The unit is cache-weighted input: fresh tokens at full weight, cache reads at a tenth. Campaign of September 22, 2026, 3 repetitions per question and arm.
14 / 30 cells · /benchmarks/2026-09-22-claude.json
Questions
How many tokens does Claude Fable 5.1 use per question in Claude Code?
In the September 22, 2026 benchmark, a bare Claude Code session on Claude Fable 5.1 sent 49,700 cache-weighted input tokens per one-line coding question on average. With capsul on the same model, 23,400.
How much does a coding question cost on Claude Fable 5.1?
At Anthropic list prices, the bare agent's traffic came to $0.636 per question and capsul's to $0.279. A Claude subscription does not bill per token: read it as how much of a plan each question uses, compared with other models.
Is the saving on Claude Fable 5.1 established?
Yes. The 90% interval runs from x1.72 to x2.65, entirely above one, on 14 measured cells.
Does capsul change Claude Fable 5.1's answers?
It passed more of them: 3 of five answer checks bare, 4 of five with capsul. Five checks is a small sample: read it as no loss rather than a gain.