CCTQ Review: Pricing & Comparison
A New API panel that publishes per-group measured availability (83.6% average) and a no-questions refund policy
Last verified: 2026-10-15 · Visit official site →
A relay that publishes how unstable it is
CCTQ (cctq.ai) runs the open-source New API panel, and its homepage copy is largely the unmodified New API template — its own About page says the administrator has not configured any about content, which is itself a candid statement about operator maturity. Where it does better than most peers is in publishing measured per-group data on its monitoring page.
What that page shows:
Overall: 4 online / 0 offline, 83.6% average availability
- ClaudeCode-MAX: 81.4% availability, 5.23s latency, 98.4% success
- CodeX-only: 100% availability, 10.17s latency
- Codex-Plus: 99.7% availability, 2.04s latency
Group exchange rates (CNY/USD): ClaudeCode-MAX 1.5, CC-Kiro-reverse 0.4, CodeX-only 0.3, Codex-Plus 0.2, image2 0.1An 83.6% average is far below anything that deserves the name SLA — and the site makes no 99.9% claim at all. Publishing unflattering numbers is genuinely rare among cheap relays and deserves credit, but it also tells you plainly: this is a service that treats instability as normal operations.
Pricing runs in ¥ per 1M tokens (claude-fable-5 at ¥15 in / ¥75 out / ¥1.5 cached; claude-opus-4-8 at ¥7.5 in / ¥37.5 out), top-ups start at ¥1, and it explicitly offers a standing no-questions refund — refund = amount topped up minus amount consumed, processed within 48 business hours, no fee — the clearest refund terms among the vendors added in this round.
Two things you must know
Mainland access is not reliable. The official docs state that API calls are not region-restricted, but that some mainland regions get blocked and users should use a proxy in that case; the token-creation page likewise notes that if you only see a group flagged as outside the service region, a proxy is required.
A reverse-engineered channel is on the menu. One public monitoring group is named CC-Kiro-reverse, a reverse-engineered, unofficial Claude source — its cheapness comes precisely from being non-compliant in origin, and the reliability and account risk sits with the user. The same logic applies to the 0.1–0.3 CNY/USD groups: very cheap rates generally map to worse stability.
Who it’s for
If you route small, non-critical, retryable traffic here and use the no-questions refund plus the public monitoring to validate, that is a reasonable low-risk pattern. For production, a published 83.6% availability and the presence of a reverse-engineered channel both argue against making it your primary. The operator is anonymous — the panel is untouched boilerplate and no company details are published — which should factor into the decision too.
Information verified 2026-10-15. Availability, latency, group rates and the refund policy reflect what cctq.ai, its monitoring page and the doc.cctq.ai documentation displayed on that date.
Related reviews
- CatRouter: a mainland-direct aggregator at ¥1 = $1 with published group multipliers, one key spanning OpenAI/Anthropic/Gemini native protocols
- APINebula: publishes input/output/cache prices and a discount ratio per model, up to 15% commission, balance never expires
- PoloAPI: balances stability and multi-model coverage, focused on usage statistics and multi-project call auditing
- PackyCode: an early domestic relay optimized specifically for Claude Code, an active community, a ¥1 new-user bonus, a top pick for coding workflows
Quick facts
| Pricing model | Billed in ¥ per 1M tokens (claude-fable-5 ¥15 in / ¥75 out / ¥1.5 cached; claude-opus-4-8 ¥7.5 in / ¥37.5 out); group exchange rates are published: ClaudeCode-MAX at 1.5 CNY/USD, CC-Kiro-reverse 0.4, CodeX-only 0.3, Codex-Plus 0.2, image2 0.1; top-ups from ¥1; a standing no-questions refund policy, refund amount = amount topped up minus amount consumed, processed within 48 business hours with no fee |
|---|---|
| Model coverage | The site claims 50+ upstream services and 100+ models with billing support; the first page of pricing renders 20 models, overwhelmingly Claude variants (fable-5/-5-1, opus-4-5 through opus-5-5, sonnet-4-5 through sonnet-5-5, haiku-4-5/5-5), plus gpt-5.6-luna/sol/terra, codex-auto-review, deepseek-v4.1-flash and glm-5.3 |
| Latency / SLA | The public monitoring page publishes per-group measurements: 83.6% average availability overall (4 online, 0 offline), ClaudeCode-MAX at 81.4% availability / 5.23s latency / 98.4% success, CodeX-only at 100% / 10.17s, Codex-Plus at 99.7% / 2.04s. The site makes no 99.9% SLA claim of any kind — a published 83.6% average is far below anything that could be called an SLA |
| Mainland direct connect | Not reliably mainland-direct: the official docs state that some mainland regions get blocked and advise using a proxy when that happens |
| Best for | Developers |
| Referral program | AFF referral reward: once an invitee's top-up has been held for 3 days, the referrer receives 5% of the recharge amount, withdrawable to balance; there is also a business-development channel for downstream Claude distribution. |
Pros
- Rare transparency: the monitoring page publishes measured per-group availability and latency (83.6% average), says outright that 90% of unavailability errors come from choosing the wrong group, and the docs admit that some mainland regions get blocked
- The clearest refund policy of the vendors added here: no-questions, no fee, refund = top-up minus consumption, processed within 48 business hours
- A published 5% referral rebate on recharges, plus self-service invoicing from ¥50
Cons
- One of its own public groups is named CC-Kiro-reverse — a reverse-engineered, unofficial Claude source, so the reliability, account and compliance risk sits with the user
- Published average availability is just 83.6%, and the cheapest groups run at 0.1–0.3 CNY/USD — the discount corresponds to visibly worse stability
- The operator is anonymous: the panel is an unmodified New API install and the About page literally says no about content has been configured
Compare more AI API relays
See the full comparison board — filter by price tier, model coverage, and mainland direct-connect status.
Back to the comparison board →