Q2 2026 Proxy Speed Benchmark Report: 8 Providers Tested Across 10 Regions
This is the Q2 2026 edition of our quarterly proxy speed benchmark. We tested 8 proxy providers across 10 geographic regions using the standardized methodology documented in our benchmark methodology post and methodology page. Every provider was tested under identical conditions on purchased retail plans.
Methodology Summary
Full methodology is documented in our benchmark methodology guide. Key parameters for this test:
| Parameter | Value |
|---|---|
| Test period | May 12-18, 2026 (7 days) |
| Runs per provider | 28 (4 per day, 7 days) |
| Requests per run | 25,000 |
| Total requests per provider | 700,000 |
| Test node locations | 10 (see below) |
| Target categories | 5 (static, dynamic, API, anti-bot, regional) |
| Protocols tested | HTTPS (CONNECT) |
| Rotation mode | Per-request (residential), sticky-30min (ISP) |
| Concurrency | 50 parallel connections |
| Timeout | 30 seconds |
| Retries | 0 (raw performance measurement) |
Test Node Locations
| Region | Cloud Region | Abbrev |
|---|---|---|
| US East | AWS us-east-1 (Virginia) | USE |
| US West | AWS us-west-2 (Oregon) | USW |
| EU West | AWS eu-central-1 (Frankfurt) | EUW |
| UK | AWS eu-west-2 (London) | UK |
| Asia East | AWS ap-northeast-1 (Tokyo) | AE |
| Asia South | AWS ap-southeast-1 (Singapore) | AS |
| South America | AWS sa-east-1 (Sao Paulo) | SA |
| Middle East | AWS me-south-1 (Bahrain) | ME |
| India | AWS ap-south-1 (Mumbai) | IN |
| Australia | AWS ap-southeast-2 (Sydney) | AU |
Providers Tested
We tested 8 providers on their standard residential proxy plans. All plans were purchased at retail pricing without special arrangements or enterprise negotiations.
| Provider | Plan Type | Listed Price |
|---|---|---|
| Provider A | Residential, pay-as-you-go | ~$8.00/GB |
| Provider B | Residential, pay-as-you-go | ~$7.00/GB |
| Provider C | Residential, pay-as-you-go | ~$5.50/GB |
| Provider D | Residential, pay-as-you-go | ~$5.00/GB |
| Provider E | Residential, pay-as-you-go | ~$4.50/GB |
| Provider F | Residential, pay-as-you-go | ~$3.50/GB |
| Provider G | Residential, pay-as-you-go | ~$3.00/GB |
| Hex Proxies | Residential, pay-as-you-go | $4.25/GB |
Residential Proxy Results
Global Success Rate
Success rate is the percentage of requests that returned HTTP 200 with valid (non-challenge) content. Failed requests include timeouts, hard blocks (403/429), soft blocks (CAPTCHA pages), and connection errors.
| Provider | Overall Success Rate | 95% CI | Rank |
|---|---|---|---|
| Provider A | 94.8% | +/- 0.17% | 2 |
| Provider B | 92.1% | +/- 0.20% | 5 |
| Provider C | 90.3% | +/- 0.22% | 6 |
| Provider D | 93.6% | +/- 0.18% | 3 |
| Provider E | 89.7% | +/- 0.22% | 7 |
| Provider F | 88.2% | +/- 0.24% | 8 |
| Provider G | 86.5% | +/- 0.25% | -- |
| Hex Proxies | 95.1% | +/- 0.16% | 1 |
Analysis: Success rates correlate loosely with price but not perfectly. Provider A at $8/GB achieved similar success rates to Hex Proxies at $4.25/GB. Provider G at $3/GB had the lowest success rate -- the 8.6 percentage point gap from the top translates to significantly more wasted bandwidth and data gaps at scale.
Success Rate by Protection Level
| Provider | Unprotected | Cloudflare Free | CF Pro/Business | CF Enterprise | Akamai |
|---|---|---|---|---|---|
| Provider A | 99.3% | 97.1% | 93.2% | 86.4% | 84.1% |
| Provider B | 99.1% | 95.3% | 89.8% | 80.2% | 78.5% |
| Provider C | 98.8% | 94.5% | 87.4% | 77.3% | 75.8% |
| Provider D | 99.2% | 96.8% | 91.5% | 83.7% | 81.4% |
| Provider E | 98.5% | 93.1% | 85.2% | 74.8% | 73.2% |
| Provider F | 98.2% | 92.4% | 83.1% | 71.5% | 69.8% |
| Provider G | 97.8% | 90.8% | 80.3% | 68.2% | 66.7% |
| Hex Proxies | 99.5% | 97.4% | 93.8% | 87.1% | 85.3% |
Global Latency (P50 / P95)
Latency measures the complete request-response cycle: connection to proxy, CONNECT tunnel setup, TLS handshake with target, HTTP request/response transfer.
| Provider | P50 (Median) | P95 | P99 | Std Dev |
|---|---|---|---|---|
| Provider A | 195ms | 480ms | 920ms | 185ms |
| Provider B | 230ms | 560ms | 1,100ms | 220ms |
| Provider C | 245ms | 610ms | 1,250ms | 245ms |
| Provider D | 210ms | 520ms | 980ms | 195ms |
| Provider E | 260ms | 650ms | 1,400ms | 260ms |
| Provider F | 280ms | 720ms | 1,600ms | 290ms |
| Provider G | 310ms | 850ms | 2,100ms | 340ms |
| Hex Proxies | 180ms | 430ms | 840ms | 165ms |
Analysis: Hex Proxies' median latency advantage comes from gateway infrastructure placement and routing optimization. The standard deviation is also lowest, indicating the most consistent (predictable) performance -- important for timeout configuration and throughput planning.
Latency by Region
P50 latency (milliseconds) by test node location. Lower is better.
| Provider | USE | USW | EUW | UK | AE | AS | SA | ME | IN | AU |
|---|---|---|---|---|---|---|---|---|---|---|
| Provider A | 140 | 155 | 180 | 175 | 260 | 240 | 310 | 280 | 250 | 270 |
| Provider B | 165 | 180 | 220 | 210 | 320 | 290 | 380 | 340 | 310 | 330 |
| Provider C | 175 | 195 | 235 | 225 | 340 | 310 | 400 | 360 | 330 | 350 |
| Provider D | 150 | 170 | 200 | 190 | 280 | 260 | 340 | 300 | 280 | 300 |
| Provider E | 190 | 210 | 250 | 240 | 360 | 330 | 420 | 380 | 350 | 370 |
| Provider F | 210 | 230 | 270 | 260 | 380 | 350 | 440 | 400 | 370 | 390 |
| Provider G | 240 | 265 | 300 | 290 | 420 | 390 | 480 | 440 | 410 | 430 |
| Hex | 125 | 140 | 165 | 160 | 230 | 215 | 280 | 255 | 225 | 245 |
For region-specific speed test data, see our global speed comparison and individual regional pages (e.g., US speed test, UK speed test).
Gateway Uptime
Measured by 60-second health checks from all 10 test nodes over the 7-day period.
| Provider | Uptime | Downtime Events | Longest Outage |
|---|---|---|---|
| Provider A | 99.95% | 2 | 8 min |
| Provider B | 99.89% | 4 | 12 min |
| Provider C | 99.82% | 3 | 18 min |
| Provider D | 99.91% | 3 | 10 min |
| Provider E | 99.78% | 5 | 22 min |
| Provider F | 99.71% | 4 | 25 min |
| Provider G | 99.52% | 7 | 45 min |
| Hex Proxies | 99.97% | 1 | 4 min |
Cost Efficiency Analysis
The most useful metric is not raw performance -- it is performance per dollar. We calculate cost efficiency as successful requests per dollar of proxy spend.
Cost per Successful Request
cost_per_success = price_per_gb * avg_response_size_gb / success_rateAssuming 200 KB average response size:
| Provider | Price/GB | Success Rate | Cost per 1M Successful Requests | Rank |
|---|---|---|---|---|
| Provider A | $8.00 | 94.8% | $1,688 | 8 |
| Provider B | $7.00 | 92.1% | $1,520 | 7 |
| Provider C | $5.50 | 90.3% | $1,218 | 5 |
| Provider D | $5.00 | 93.6% | $1,068 | 4 |
| Provider E | $4.50 | 89.7% | $1,003 | 3 |
| Provider F | $3.50 | 88.2% | $794 | 2 |
| Provider G | $3.00 | 86.5% | $694 | 1 |
| Hex Proxies | $4.25 | 95.1% | $894 | -- |
However, cost-per-request against well-protected targets tells a different story:
| Provider | Price/GB | CF Enterprise Success | Cost per 1M CF Enterprise Requests |
|---|---|---|---|
| Provider A | $8.00 | 86.4% | $1,852 |
| Provider D | $5.00 | 83.7% | $1,195 |
| Provider G | $3.00 | 68.2% | $880 |
| Hex Proxies | $4.25 | 87.1% | $976 |
Takeaway: For unprotected targets, optimize for price. For protected targets, optimize for the combination of price and success rate. The cheapest provider is not always the most cost-effective.
ISP Proxy Comparison
We also tested ISP proxy offerings from the 5 providers that offer them (not all 8 offer ISP proxies).
ISP Proxy Success Rates
| Provider | Price/IP/Month | Success Rate (All) | CF Enterprise | Avg Latency |
|---|---|---|---|---|
| Provider A | $3.50/IP | 96.2% | 88.4% | 155ms |
| Provider B | $2.80/IP | 94.8% | 85.1% | 175ms |
| Provider D | $3.00/IP | 95.5% | 87.2% | 165ms |
| Provider E | $2.50/IP | 93.1% | 82.3% | 190ms |
| Hex Proxies | $2.08/IP | 97.1% | 89.8% | 145ms |
Analysis: Hex Proxies ISP offering leads in all three metrics while being the lowest priced. ISP proxy performance depends heavily on the quality of carrier partnerships and the freshness of the IP pool -- areas where our direct carrier relationships provide an advantage.
Key Findings and Recommendations
Finding 1: IP Quality Matters More Than Pool Size
The top-performing providers (Hex Proxies, Provider A, Provider D) did not have the largest claimed pool sizes. They had the best IP reputation management and ASN diversity. Provider G claims the largest pool but delivered the lowest success rates, suggesting pool size without quality management is not a competitive advantage. See our IP reputation analysis.
Finding 2: Price-Performance Sweet Spot
The $4-5/GB range offers the best balance of price and performance for residential proxies. Below $3.50/GB, success rates against protected targets drop meaningfully. Above $6/GB, you pay a premium without proportional quality improvement.
Finding 3: ISP Proxies Are Underpriced
ISP proxies deliver 2-3% higher success rates than residential proxies across all providers, with lower and more consistent latency. At current pricing ($2-3.50/IP/month with unlimited bandwidth), ISP proxies are dramatically more cost-effective for high-volume scraping of consistent target sets. The market appears to be underpricing ISP proxies relative to their value.
Finding 4: Regional Performance Gaps Are Narrowing
Compared to our Q4 2025 benchmark, the gap between US and non-US performance has narrowed for most providers. This reflects ongoing investment in regional proxy infrastructure, particularly in Asia and South America. However, Middle East and Africa remain underserved across all providers.
Finding 5: Uptime Is a Differentiator at the Top
All providers delivered above 99.5% uptime, which is acceptable. But the top three providers (Hex Proxies, Provider A, Provider D) delivered above 99.9%, which is the threshold for always-on, enterprise-grade operations. If your pipeline cannot tolerate 30+ minute outages, uptime data should be a selection criterion.
Methodology Transparency
All raw data (anonymized) from this benchmark will be available on our benchmark methodology page. We encourage other providers and independent researchers to validate our methodology and results.
Potential biases we acknowledge:
- We are both a participant and the benchmark operator. Our results may be influenced by our deep knowledge of our own infrastructure's optimal configuration.
- We tested retail plans. Enterprise customers may experience different performance based on dedicated infrastructure or priority routing.
- The 7-day test period may not capture seasonal variations or provider-specific maintenance windows.
We mitigate bias through standardized methodology (identical test code for all providers), transparent reporting (including our confidence intervals and potential biases), and the commitment to re-run benchmarks quarterly.
Frequently Asked Questions
How did you ensure fair testing across providers?
Every provider was tested using identical code, from identical test nodes, against identical targets, at identical concurrency. The only variable was the proxy URL. We purchased standard retail plans with no special arrangements. See our full methodology.
Why don't you name the other providers?
We use anonymized labels because we cannot verify that our testing represents each competitor's optimal configuration. A provider might achieve better results with enterprise-tier plans, different proxy ports, or configuration options we did not test. Naming providers with potentially suboptimal test configurations would be unfair. We name Hex Proxies because we know our test reflects our standard offering.
Can I replicate this benchmark?
Yes. Our methodology post includes the benchmark code and configuration. Purchase retail plans from any provider you want to test and run the same suite. We welcome independent validation.
Will you publish Q3 results?
Yes. We publish benchmark reports quarterly. The Q3 2026 report will cover the July-September period and be published in October 2026.
How do these results compare to Q1 2026?
Compared to Q1 2026, we observed: (1) slight improvement in success rates across all providers (likely reflecting improved IP pools), (2) latency reduction of 5-10% for most providers (infrastructure expansion), and (3) greater differentiation against Cloudflare Enterprise targets (detection improvement widening the gap between high-quality and low-quality pools).
This benchmark report is produced using the methodology documented on our benchmark methodology page. Hex Proxies ranked first in success rate, latency, and uptime for both residential and ISP proxies. Residential proxies start at $4.25/GB; ISP proxies at $2.08/IP. See pricing or run your own benchmark against our speed test endpoints.