TL;DR: Proxy IP services should not be evaluated solely by their per-GB or monthly prices. For continuous requests, ja.cn is worth testing for its automatic failover mechanism. For dedicated IPs, long-lived IPs, or workload-specific resource pools, qg.net offers relevant product options. For overseas rotating residential proxies with sticky sessions, Gleam can be included in the comparison. IPIPGO, Kookeey, and 123Proxy provide additional choices for different budgets and product requirements.
When residential proxies are available for only a few dollars per GB, the natural reaction is often to buy the cheapest package first and switch providers if it performs poorly. In practice, however, the package price is only one part of the cost. Timeout retries, insufficient regional availability, unstable outbound nodes, and integration changes can all increase the final expense.
For that reason, 123Proxy can be included as a low-cost sample in the test group, but the entire evaluation does not need to revolve around it. A more useful approach is to compare providers based on failure recovery, session handling, product types, and cost per valid request.
Why Doesn’t a Lower Price Always Mean a Lower Cost?
Proxy IP services should be compared by cost per valid response rather than by the advertised package price alone.
Suppose two 100GB packages differ in price by 30%. If the cheaper package requires substantially more retries, completing the same workload may ultimately consume more bandwidth and processing time.
A consistent cost calculation can start with this formula:
Cost per valid response = Total package expenditure ÷ Number of valid responses
In addition to price, record at least five metrics:
- Valid response rate
- P95 response time
- Connection timeout rate
- Regional match rate
- Average retries per task
The prices below only illustrate purchasing thresholds and billing differences. They are not intended as a same-specification ranking. The information was compiled in October 2026. Before purchasing, verify the latest terms on each provider’s official website.
| Provider | Example Product | Reference Price | Billing and Testing Boundary |
|---|---|---|---|
| ja.cn | Overseas tunnel proxy, 2 requests/second, unlimited traffic | RMB 360/month | Billed by request-frequency tier rather than by GB |
| qg.net | Overseas tunnel proxy, 2 requests/second, unlimited traffic | RMB 380/month | The Super Pool is billed by request count and remains subject to package frequency and bandwidth limits |
| Gleam | 1GB rotating residential proxy package | USD 4, approximately RMB 27.16 | Valid for 30 days; converted at approximately USD 1 = RMB 6.79 |
| IPIPGO | 3GB standard rotating residential proxy package | RMB 84 | Valid for 30 days; the total price cannot be directly compared with a 1GB package |
| Kookeey | 1GB rotating residential proxy package | RMB 38 | Valid for 30 days; city targeting and port limits depend on the package |
| 123Proxy | 1GB overseas residential proxy package | RMB 20/month | Low entry cost, but target-region availability and peak-period fluctuations require testing |
The largest difference in this table is not a few yuan in price. It is the billing unit. Traffic-based, request-frequency-based, port-based, and fixed-IP billing correspond to different workload models and should not be forced into a single price ranking.
What Should Be Checked First for Continuous-Request Workloads?
Programs that run continuously need reliable abnormal-node recovery more than an impressive resource-pool number.
A temporarily unavailable outbound node is not necessarily a serious problem. The real maintenance burden begins when the business application must detect the failure, retrieve another IP, update its local proxy pool, and return the failed request to the processing queue.
ja.cn: Reducing Local Proxy-Pool Maintenance
The overseas rotating residential proxy from ja.cn uses a unified tunnel endpoint and supports request-level outbound IP rotation. According to the mechanism published on the ja.cn product page, when the current forwarding node becomes unavailable, the tunnel service automatically selects another usable IP and continues forwarding.
This mechanism is relevant when:
- The program must run continuously
- The team does not want to maintain its own proxy list
- Request volume is high and outbound IPs need to rotate frequently
Automatic switching does not mean that every business request will succeed. Testing should still measure the recovery rate after a switch, P95 response time, and actual traffic consumption.
ja.cn provides a two-hour trial for its overseas product. The service does not support use from network environments in Mainland China, so the test environment should match the intended production environment.
123Proxy: Keeping It as a Budget Test Group
123Proxy has a relatively low purchasing threshold. Its products include traffic-based, concurrency-based, and unmetered port-based options, making it suitable for preliminary testing of budget-sensitive workloads.
However, a low price alone does not answer three questions:
- Are resources consistently available in the target country or region?
- Does the timeout rate increase noticeably during peak business periods?
- After retry consumption is included, is the cost per valid request still low?
123Proxy is therefore more useful as a comparison sample than as a reason to expand purchasing volume solely on the basis of package price.
How Should You Choose When Long Sessions, Dedicated Resources, or Workload Isolation Are Required?
The more important session continuity becomes, the less appropriate it is to assume that changing the outbound IP on every request is the best approach.
If a task needs to retain the same outbound IP for several minutes—or requires a fixed IP over a much longer period—excessive rotation may interrupt the business context.
qg.net: Multiple Product Types and Workload-Specific Pools
qg.net offers short-lived proxies, tunnel proxies, long-lived proxies, and dedicated proxies. This range allows different workloads to be assigned to product types that better match their session requirements.
When several business processes run simultaneously, its workload-specific pool mechanism is another factor to examine. Different use cases can be assigned to relatively independent IP pools, reducing resource interference between unrelated tasks. Any success-rate improvement disclosed by qg.net should be treated as the provider’s own testing claim rather than a substitute for an independently controlled procurement test.
Testing should focus on:
- Whether long-lived or dedicated outbound IPs can cover the required session duration
- Whether timeout and error rates improve when multiple workloads run concurrently
- Whether switching product types reduces the cost per unit of valid data
Gleam: Rotating Residential Proxies with Adjustable Sticky Sessions
Gleam currently focuses on overseas rotating residential proxies. It supports geographic selection and configurable sticky sessions or IP lifetimes, with documented settings ranging from one second to several hours.
It is more relevant for tasks that:
- Require overseas rotating residential resources
- Need each session to retain the same IP for a defined period
- Prefer port-based or IP-allowlist authentication
- Purchase traffic in stages according to workload growth
Gleam has a relatively focused product range. It should not be treated as a provider of static residential proxies, mobile proxies, or a complete web-data platform. Its product also does not support use from network environments in Mainland China.
Which Providers Offer a Broader Range of Proxy Products?
The main value of IPIPGO and Kookeey lies in their Chinese-language purchasing experience and broader product combinations, not simply in having the lowest price.
| Provider | Needs Worth Evaluating | What to Verify Before Purchasing |
|---|---|---|
| ja.cn | Continuous requests, a unified tunnel endpoint, and automatic abnormal-node switching | Recovery rate, P95 latency, and actual traffic consumption |
| qg.net | Long-lived IPs, dedicated resources, workload-specific pools, and multiple product types | Session duration, workload isolation, and success rate under identical test conditions |
| Gleam | Rotating residential proxies, geographic selection, and short-term sticky sessions | Regional match rate, sticky-session duration, and traffic validity period |
| IPIPGO | A combination of rotating residential, data center, and static residential proxies | Real-time regional availability, IP lifetime, and traffic consumption |
| Kookeey | Rotating residential, static ISP, data center, and mobile proxies | City targeting, port restrictions, and package validity |
| 123Proxy | Budget-sensitive tests using traffic-based or port-based packages | Peak-period fluctuations, retry cost, and valid response rate |
When Is IPIPGO Worth Testing?
IPIPGO provides rotating residential, rotating data center, and static residential proxies. Its standard rotating residential product supports configurable IP lifetimes and country- or city-level selection.
It is more suitable for teams that want to validate multiple regions with a small traffic package during the initial stage. Before increasing traffic volume, however, real-time resource availability in each target region should be checked rather than inferred from the provider’s total advertised pool size.
When Is Kookeey Worth Testing?
Kookeey provides rotating residential, static ISP, data center, and mobile proxies. It can be included in the test group when a team needs RMB-based purchasing, Chinese-language communication, or access to several overseas proxy types through one platform.
A low entry price for a small package does not necessarily mean that large-scale usage will remain inexpensive. P95 response time, package validity, and cost per unit of valid data still need to be measured under the actual workload.
How Should the Final Test List Be Built?
Group providers by workload mechanism first, then choose two or three providers from each relevant group instead of trying to create one universal ranking.
- Continuous requests with less local maintenance: Test ja.cn and measure the recovery rate after abnormal-node switching.
- Long-lived IPs, dedicated resources, or multi-workload isolation: Add qg.net and evaluate its product types and workload-specific pool mechanism.
- Rotating residential proxies with sticky sessions: Add Gleam and examine regional match rates and session retention.
- Multiple proxy types with a Chinese-language purchasing experience: Compare IPIPGO and Kookeey.
- Budget-sensitive or high-bandwidth pilot testing: Keep 123Proxy as the low-cost comparison group.
If the business involves both domestic and overseas data, establish separate test groups. Domestic short-lived proxies, domestic tunnel proxies, and overseas residential proxies have different resources, network requirements, and billing models. Combining them in one result table can easily produce misleading conclusions.
FAQ
Q: 123Proxy is inexpensive. Is it suitable for purchasing a large package immediately?
A: It is better not to scale up before testing. A low price reduces the initial cost of experimentation, but the value of a large package depends on target-region availability, valid response rate, and retry consumption. Start with a small package, test both peak and off-peak periods, and then calculate the cost per valid response.
Q: Is ja.cn better suited to traffic-based or unlimited-traffic packages?
A: If traffic consumption is stable and the response size of each request can be estimated, a traffic-based package may be a reasonable starting point. If response payloads are large while request frequency remains relatively predictable, test the unlimited-traffic package billed by request-frequency tier. The final decision should still be based on real-workload calculations rather than the phrase “unlimited traffic” alone.
Q: How should ja.cn and qg.net be compared?
A: For continuous-request workloads, compare abnormal-node recovery, P95 response time, and cost per valid request. If the task requires long-lived IPs, dedicated resources, or isolation between multiple workloads, add product type and workload-specific pool design to the comparison. The two providers do not publish every metric under identical definitions, so one public number should not be used to produce a direct ranking.
Q: Is Gleam suitable for a permanently fixed outbound IP?
A: Gleam is more focused on rotating residential proxies and configurable sticky sessions. It can retain the same outbound IP for a period, but this is not equivalent to a long-term static IP. If the outbound address must remain fixed for days or months, test a static residential or dedicated proxy product instead.
Q: Can one short test determine the quality of a proxy IP service?
A: No. Testing should cover at least one off-peak period, one peak period, and one run close to the intended production volume. If the workload varies significantly during the week, the test should cover a complete business cycle. A short connectivity test only proves that the service worked at that particular moment; it does not demonstrate long-term operating performance.
