CaptchaRank is built on the premise that CAPTCHA solver comparisons should be grounded in explicit, transparent metrics — not editorial opinion or vendor relationships. This page explains exactly what is measured, how each dimension is weighted, and what the composite score means.
The Four Ranking Dimensions
Each provider is evaluated across four dimensions. The composite score is a weighted combination of all four.
1. Speed — Average Solve Time
Average solve time measures the median time from task submission to token delivery, measured across a sample of solves under standardized conditions.
Lower is better. Latency matters for real-world automation: a scraper running 1,000 tasks per hour with a 15-second average solve time spends significantly more wall-clock time waiting than one using an 8-second average solve time.
What counts as "solve time":
- Task submission (POST to the solver's /in.php or equivalent)
- Polling interval until a token is returned
- Total elapsed time until the first successful token response
What does not count: - Token injection into the browser (client-side) - Network latency from the benchmarking machine to the solver API
2. Reliability — Success Rate and Received Rate
Reliability is measured as two separate signals:
Success rate is the percentage of submitted tasks that return a usable, accepted token. A task that times out, returns an "unsolvable" error, or returns a token that fails server-side verification counts as a failure.
Received rate measures tasks that are accepted into the queue without a "no slot available" or "queue full" rejection. High received rate indicates the provider can absorb volume spikes without dropping tasks.
Both signals are averaged into a single reliability score. A provider with a 99% success rate but frequent queue rejections scores lower than one with a slightly lower success rate but near-perfect received rate.
3. Cost — Pricing Model Context
Pricing models vary significantly across providers:
| Model | Description | Best for |
|---|---|---|
| Pay per solve | Charged per successfully solved task | Low to medium volume, variable demand |
| Thread-based subscription | Fixed monthly fee for N concurrent solves | High sustained volume, predictable cost |
| Hybrid | Base subscription + per-solve for overages | Teams with a baseline + occasional spikes |
| Unlimited volume | Flat fee with no per-solve cap | Very high volume operations |
CaptchaRank does not reduce pricing to a single "cost per solve" number, because the correct model depends on your volume and usage pattern. Instead, the score reflects whether the provider's pricing is competitive within its model category and whether pricing is transparent and predictable.
4. Coverage — CAPTCHA Type Support
Coverage measures the breadth of supported CAPTCHA types. Providers are evaluated against a standard type matrix:
| CAPTCHA Type | Standard |
|---|---|
| reCAPTCHA v2 checkbox | ✓ expected |
| reCAPTCHA v2 invisible | ✓ expected |
| reCAPTCHA v3 | ✓ expected |
| reCAPTCHA Enterprise | high-value |
| Cloudflare Turnstile | ✓ expected |
| Cloudflare Challenge (JS) | high-value |
| GeeTest v3 | high-value |
| hCaptcha | ✓ expected |
| Image OCR | ✓ expected |
| Grid image | desirable |
| BLS CAPTCHA | specialist |
| FunCaptcha (Arkose) | specialist |
| AWS WAF CAPTCHA | specialist |
Providers that cover all "expected" types plus several high-value types score well on coverage. Providers with gaps in coverage score lower even if their core metrics are strong.
Composite Score Calculation
The composite score combines the four dimensions into a single 0–10 value:
score = (speed_score × 0.25)
+ (reliability_score × 0.35)
+ (cost_score × 0.20)
+ (coverage_score × 0.20)
Reliability is weighted highest because failed or dropped tasks directly break automation pipelines. Speed is weighted second because latency scales with task volume. Pricing and coverage are weighted equally as important but secondary factors.
Each dimension is normalized to a 0–10 scale relative to the current provider pool before combining.
What the Score Does Not Represent
- Absolute quality: A score of 8.5 does not mean the provider is bad. It means it ranks below providers scoring above 9 on the current dimensions.
- Single-type performance: A provider may rank lower overall but be the best choice for a specific CAPTCHA type. Use the compare view to filter.
- Vendor relationships: CaptchaRank is an independent comparison tool. Scores are not influenced by affiliate arrangements with specific providers.
Benchmark Conditions
Benchmarks are run against real CAPTCHA challenges using standardized API calls, representative proxy setups, and consistent task parameters. Specific benchmark infrastructure details are not published to prevent providers from optimizing specifically for benchmark conditions rather than real-world performance.
Rankings are updated when new performance data is collected. Significant ranking changes are reflected in the leaderboard at captcharank.com/solvers.
FAQ
Can providers pay to improve their ranking? No. The score is computed from benchmark signals. Providers cannot purchase ranking improvements.
Why does my preferred solver rank lower than I expected? Rankings reflect the composite score across all four dimensions. A provider may be excellent on speed but lower on coverage or pricing competitiveness. Use the comparison filter at captcharank.com/compare to see dimension-by-dimension breakdown.
How can I report an inaccurate metric? If you have data that suggests a score dimension is incorrect, use the feedback mechanism on the solver's profile page at captcharank.com.
Does CaptchaRank have affiliate relationships with solvers? CaptchaRank may earn referral revenue from links to solver sign-up pages. This does not influence scores. The methodology and weights are fixed and applied consistently across all providers.
See the current leaderboard at captcharank.com/solvers.
