Tencent Cloud Token Plan: No Free Tier, Starts at ¥28/mo — General vs Hy Pricing Compared

Tencent Cloud Token Plan is the most unusual AI coding subscription on the market right now — it has no free tier, but its ¥28/mo entry price is among the lowest of any mainstream Coding Plan. And it's split into four completely different product lines where picking the wrong one isn't just a price issue — your model library and integration method change entirely.

This article breaks down pricing for all four lines based on official Tencent Cloud docs — no marketing fluff, just the numbers. Prices below are list prices as of 2026-08-12. Check the Tencent Cloud website for the latest rates.

Is there a free tier? No.

Tencent Cloud Token Plan currently has no free tier, no trial credits, and no first-purchase token bonuses. All four product lines — Personal General, Personal Hy, Enterprise Lite, and Enterprise Pro — are prepaid monthly subscriptions that charge immediately upon purchase, with no try-before-you-buy option.

Two things soften the "no free" situation:

1. Hy Lite at ¥28/mo is inexpensive enough to try for a month and drop if it doesn't fit — no long-term commitment.
2. You can hold General and Hy simultaneously (one of each), sharing a single API key. Start with one, add the other if needed.

Compare to Cursor at $20/mo (~¥144) or Kiro Free plus Pro at $20/mo — Tencent Cloud's ¥28/mo is genuinely the cheapest paid entry among Coding Plans.

Four lines explained: it's not just price — billing models and model pools are completely different

Tencent Cloud Token Plan's four lines aren't tiered pricing — they're fundamentally different products. General is a multi-model pool, Hy is a single-model path, Enterprise Lite is a token pool with a single router, and Enterprise Pro is a credit pool with multi-key governance. Here's how they compare (source):

General series: largest model pool, built for switching

The General series is your ticket to multi-model access. Auto smart routing distributes tasks to the best-fit model automatically, and you can also manually choose from GLM-5/5.1/5.2, Kimi-K2.5, MiniMax-M2.7/M3, DeepSeek-V4 series, and more.

Four token tiers: Lite ¥39/mo (35M), Standard ¥99/mo (100M), Pro ¥299/mo (320M, recommended), Max ¥599/mo (650M). (source)

For: Developers wanting to explore different models for Agent tasks, or those needing specific model capabilities like GLM or DeepSeek.

Hy series: Hunyuan-only, 20-30% cheaper

The Hy series logic is: "We'll give you our single best in-house model, but at a lower price." The four Hy tiers:

• Lite ¥28/mo (35M tokens)
• Standard ¥79/mo (100M tokens)
• Pro ¥238/mo (320M tokens)
• Max ¥468/mo (650M tokens)

Equivalent tiers are 20-30% cheaper than General (Hy Lite ¥28 vs General Lite ¥39, Hy Pro ¥238 vs General Pro ¥299, Hy Max ¥468 vs General Max ¥599), but the model pool is exactly one model: Hy3.

Hy3 is genuinely competitive for Agent work: 295B/21B MoE architecture, 256K context, near-frontier on SWE-bench Verified. If you trust Hunyuan, this line is the most affordable path.

The four token tiers mirror the General series exactly (35M / 100M / 320M / 650M) — just with a dramatically narrower model pool. (source)

For: Developers committed to Hunyuan Hy3 who don't need third-party model comparisons — fewer tokens cost, same model power.

General vs Hy: a decision framework

1. If one model can handle all your use cases → pick Hy, save money.
2. If you want to switch models for different tasks (coding with A, analysis with B, long-form with C) → pick General, pay 10-20% more.
3. If you're unsure → try Hy Lite at ¥28 for a month, add a General plan later if you miss the third-party models.

Enterprise lines: not "more expensive" — a completely different product architecture

Enterprise Lite and Enterprise Pro are separate from personal plans — different Base URLs, different key systems, different billing units. They can't be mixed with personal plans. (source)

Enterprise Lite: Token pool + Auto only

The simplest enterprise option: ¥2 per million tokens, self-select a monthly pool (50M–5B tokens), minimum ~¥100/month. But only Auto routing is supported — you can't pick specific models. You get one API key per 50M tokens of quota, with per-key TPM caps and total limits. (source)

For: Small teams that don't want to manage model selection, just want Auto routing for all AI coding requests, needing multiple keys for different members.

Enterprise Pro: Credit pool + 14 models + multi-key governance

Enterprise Pro is the only line that uses a credit system: 100 credits = ¥1, monthly budget ¥1,000–¥20,000 set in the console. It supports all 14 models (the full General series pool) with per-project keys, each with independent TPM and quota limits. (source)

This is the only line that truly enables "team project procurement" — an R&D team can allocate keys by project, each with its own model permissions and usage caps.

For: Multi-project R&D teams needing fine-grained quota governance and model-level access control.

Token Plan vs direct API: two separate worlds

Tencent Cloud Token Plan and Hunyuan Token API are completely separate systems — different keys, different URLs, different billing. This is the most common confusion: the regular API key you apply for in the TokenHub console is for backend pay-as-you-go usage, not for connecting Claude Code or Cursor.

Tencent claims Token Plan saves 50%+ versus direct text-generation API, but the trade-off is no backend scripts, no automation pipelines, no account sharing.

Decision: If you use Claude Code, Cursor, OpenClaw, Cline for AI coding → Token Plan. If you're building API services, running batch inference, or doing production integration → Hunyuan API pay-as-you-go.

Quick reference: which line for which situation

No free tier, but ¥28/mo is low enough to try. When choosing, don't get distracted by the price gap — first figure out how many models you need and whether you need team key allocation, then pick accordingly.