Updated November 12, 2026: this guide now distinguishes Alibaba ModelStudio Coding Plan from Token Plan Personal Edition and Qwen account subscriptions. The old supported-model list has been replaced with the current official list. The original URL and publication date are preserved.
If you searched for “Qwen Cloud Coding Plan,” start by identifying the actual product in your console. Similar model names do not make subscriptions, quotas or API endpoints interchangeable. A price comparison is only useful when it compares access you can really use.
What the $50 Coding Plan includes
Alibaba's current Coding Plan overview lists Pro at US$50/month, with simultaneous caps of 6,000 model calls per rolling five hours, 45,000 per week and 90,000 per month. The first cap reached pauses calls. These are not three pools to add together, and one coding task can make several calls.
The listed exact models are qwen3.7-plus, qwen3.6-plus, kimi-k2.5, glm-5, MiniMax-M2.5 and qwen3.5-plus. Do not assume every Qwen model is included. Lite stopped accepting new subscriptions March 20 and renewals/upgrades April 13. Pro availability is limited; the documentation describes daily restocking at midnight UTC+8.
Token Plan Personal is a different purchase
The Personal Edition overview, updated November 12, lists the following Singapore-region plans. Credits are a billing unit, not a fixed count of coding tasks.
| Plan | Published monthly price, USD | Monthly Credits |
|---|---|---|
| Lite | $6 | 20,000 |
| Essential | $10 | 45,000 |
| Standard | $18 | 80,000 |
| Pro | $43 | 200,000 |
Eligible new users have temporary Lite/Essential prices of $3/$5 through November 30 at 23:59:59 UTC+8. November 9 brought quota increases and a Pro price reduction; existing Pro pricing changes on the next subscription cycle. Unused monthly Credits do not roll over. Additional 20,000-Credit bundles are listed at $15/month, up to five. Confirm the checkout total and eligibility before buying.
A Coding Plan request is not a Token Plan Credit
Do not divide one plan's price by the other plan's quota and call the result cheaper. Request-count and Credit-count denominators measure different things. Token Plan consumption varies with the selected model and workload. Coding Plan's multiple time windows can constrain a burst of work even when the monthly allowance looks ample.
Nerova's recommendation is to compare a representative set of completed tasks. Record billed usage, time spent reviewing, failures and how often a quota interrupts the task. Keep the same repository snapshots and acceptance criteria across trials. A useful result is cost per accepted change, not the largest number in a quota table.
The endpoint mistake that can break your budget
Alibaba's setup guide says Token Plan, Coding Plan and pay-as-you-go keys and base URLs must be matched to their own products. A key prefix alone is insufficient: both plan families can use sk-sp-.
- Coding Plan OpenAI-compatible endpoint:
https://coding-intl.dashscope.aliyuncs.com/v1 - Token Plan Personal OpenAI-compatible endpoint:
https://token-plan.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1
Copy the endpoint from the subscription you actually purchased. Run one small approved test and check the provider's usage screen before a larger batch. Do not publish credentials in troubleshooting screenshots, source control or support questions.
How to choose without overbuying
- Confirm the model and tool combination works on the specific plan.
- Check allowed usage and account rules before using a personal subscription for a shared team or production service.
- Trial typical work rather than only short prompts. Measure bursts as well as monthly volume.
- Set a spending limit and stop uncontrolled retries.
- Recheck pricing after promotions expire or your subscription renews.
For Qwen account OAuth access rather than a provider API plan, see our separate Qwen Code pricing guide. Those entitlements should not be mixed into this table.
Bottom line
The $50 Coding Plan remains a request-capped option; the lower-priced Token Plan tiers use a different billing structure. Neither is unlimited infrastructure. Choose the product whose eligible models, usage rules and measured task economics fit the work, then verify that your client is charging the intended account.