How Much Opus 5.5 Does Claude Pro Include? x2v Quota Measurement
This page is an x2v first-party measurement (2026-11-04, Claude Code) of one Claude Pro account, not a quota published by Anthropic. Bottom line: in API-equivalent terms, a month of Claude Pro on Opus 5.5 is worth about $800–1,390, roughly 40–70× the $20/month price. The weekly cap, not the 5-hour window, is what binds. The original artifact is linked in Sources.
API prices for related models
Third-party models are billed at API rates on the new plan; compare channel prices on each model page.
Estimated allowance (API-equivalent)
5-hour window: output-heavy ≈1.3M output tokens, $26–29; agentic coding mix (99% cache reads, 0.2% output) ≈160M total tokens, $44.
1 week (≈7.2 five-hour windows, range 6.8–7.8): output-heavy 9.3M output tokens, $185–205; agentic 1.15B tokens, $320.
1 month (weekly cap ×4.35): output-heavy 40M output tokens, $800–890; agentic 5.0B tokens, $1,390. That is ≈40–70× the $20/month Pro price.
Measurement method
147 controlled Opus 5.5 calls, headless `claude -p --model claude-opus-5-5 --effort medium`, no tools or MCP. List prices used: $4/M input, $20/M output, $8/M 1h cache writes, $0.20/M cache reads.
Batches: A1 and A2 each wrote 20 essays (about 5,000 output tokens per call). B0–B2 were 86 calls with a fixed ~106k context and a one-sentence answer, to isolate cache reads.
Meter readings (cumulative, UTC)
Columns: output / cache write / cache read / API $ / 5h meter / weekly meter.
Start 10:26: 714 / 23.6k / 112k / $0.23 / 0% / 4%
A1 10:29: 108k / 99k / 732k / $3.11 / 11% / 5%
A2 10:31: 214k / 121k / 1.11M / $5.48 / 19% / 6%
B0+B1 10:32: 222k / 252k / 4.31M / $7.32 / 23% / 7%
B2 10:54: 237k / 306k / 11.0M / $9.39 / 28% / 7%
Fit: cache counts at roughly half price
Model: meter % = (output $ + a·cache-write $ + b·cache-read $) / L × 100. Charging everything at list price (L = $31.75) overestimates. The fit a ≈ 0.55, b ≈ 0.5, L ≈ $25.75 output-equivalent per 5 hours cuts the error by about 10×. Thinking tokens count as output.
Weekly cap ratio
Across the run the weekly meter moved 3 points while the 5-hour meter moved 28 points, so one full 5-hour window is about 1/7.2 of the week (range 6.8–7.8). The weekly cap binds.
Limitations
The meter shows integer percentages (±0.5pt); the weekly ratio rests on a few points of movement. Single day and time of day, one account. Batch B2 may have been throttled. This is a measurement, not an Anthropic commitment.
How to use this data
- Recalculate with your own output, cache-write and cache-read mix; agentic coding is mostly cache reads.
- Budget against the weekly cap: about 7.2 full five-hour windows per week.
- Single-account, single-day measurement; check your own usage meter before relying on it.
Frequently asked questions
How much Opus 5.5 does Claude Pro include?
Per x2v measurement (2026-11-04): ≈1.3M output tokens per 5-hour window and ≈9.3M per week for output-heavy work; ≈160M total tokens per 5h for agentic coding.
Is Claude Pro worth it versus the API for Opus 5.5?
A fully used month is worth about $800–1,390 at API list prices, ≈40–70× the $20 price, if you can use the weekly cap.
Is this an official Anthropic quota?
No. It is an x2v first-party measurement with Claude Code on one account and one day; Anthropic does not publish token allowances for Pro.
Official sources and review date
Billing guidance reviewed: