Claude Opus 5.5 in Claude Code: tokens per question
This newer Opus release combines the family's coding capability with a lower API list price than the earlier Opus rows.
The figures
- Bare Claude Code
- 42,300
- With capsul
- 15,900
- less input sent
- x2.66
- answer checks passed out of five, bare then with capsul
- 3 / 4
- tool steps per question, bare then with capsul
- 5.2 / 3
- output tokens per question, bare then with capsul
- 1,490 / 548
- API-equivalent cost per question, bare then with capsul
- $0.222 / $0.0696
cache-weighted input tokens per question
cache-weighted input tokens per question
90% interval: x1.81 to x3.92
Its cache-read price is also lower than those earlier rows, so the mix of fresh and cached input matters when you estimate an API-equivalent cost. For a subscription user, that estimate is a way to compare model traffic, not a charge added to the plan. Difficult debugging and multi-file changes are sensible reasons to consider an Opus model.
The measured questions are short and repeatable. Read the interval, the answer checks and the individual questions together. They tell you more about this protocol than a single headline multiplier and leave room for a long session to behave differently.
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 | 51,800 | 22,100 | x2.35 | 0 / 0 |
| Q2 | 38,300 | 16,400 | x2.33 | 2 / 3 |
| Q3 | 34,900 | 10,300 | x3.37 | 3 / 3 |
| Q4 | 51,400 | 18,300 | x2.81 | 1 / 3 |
| Q5 | 35,300 | 12,400 | x2.85 | 3 / 3 |
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.
30 / 30 cells · /benchmarks/2026-09-22-claude.json
Questions
How many tokens does Claude Opus 5.5 use per question in Claude Code?
In the September 22, 2026 benchmark, a bare Claude Code session on Claude Opus 5.5 sent 42,300 cache-weighted input tokens per one-line coding question on average. With capsul on the same model, 15,900.
How much does a coding question cost on Claude Opus 5.5?
At Anthropic list prices, the bare agent's traffic came to $0.222 per question and capsul's to $0.0696. 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 Opus 5.5 established?
Yes. The 90% interval runs from x1.81 to x3.92, entirely above one, on 30 measured cells.
Does capsul change Claude Opus 5.5'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.