What is the most frustrating thing about buying an AI coding plan now?

It is not that you cannot find one, but that there are too many choices.

When you see a cheap one, you worry the model may not be suitable; when you see one with a very large quota, you do not know whether you can use it all up. After finally choosing one, you still have to confirm: can it connect to the tools you usually use? If you switch to another model, do you have to open another account and buy another plan?

If you are also hesitating back and forth between these questions, you can take a look at AceDataCloud Coding Plan: the page shows prices starting from ¥99, it can connect to coding tools such as Claude Code and Codex, while covering multiple model ecosystems.

However, do not rush to place an order just because of “¥99.”

What exactly does it buy? Is it suitable for you? These questions are more worth discussing than the price itself.

Does ¥99 buy official memberships for Claude Code and Codex?

No. This distinction must be made clear.

What you buy here is AceDataCloud's model invocation plan, and then, according to the corresponding guide, connect it to clients such as Claude Code and Codex. It does not open official subscriptions from both companies for you, nor does it mean you receive all benefits of official memberships.

Does that sound a little complicated? Actually, it can be broken down into two things:

What tool you are accustomed to using for coding, and what model that tool calls behind the scenes.

The former is a work habit, and the latter is a model choice.

Some people like using Claude Code in the terminal, while others are more accustomed to Codex. Just because you want to switch models, do you necessarily have to switch development tools too?

The appeal of plans like Ace lies here: you can start with the tools you are familiar with, and then select available models according to the configuration guide. Its official model coverage description includes ecosystems such as Claude, GPT/Codex, GLM, Kimi, and DeepSeek; specific versions and combinations are subject to the real-time catalog and the client compatibility scope.

Of course, “multiple models available” does not mean every tool can call every model without distinction. Confirming the combination you want to use before purchase is always more comfortable than troubleshooting after buying.

How should you choose by price? Instead, I do not recommend buying an expensive one at the beginning

First, lay out the numbers.

The official page lists the following four plan tiers, with a 30-day period. Only base quotas are listed here; limited-time bonuses are not counted as long-term benefits.

Plan Displayed price／30 days Base quota
Coding Plus ¥99 400 Credits
Coding Standard ¥299 1,200 Credits
Coding Pro ¥699 3,000 Credits
Coding Max ¥1,399 8,000 Credits

Actual prices and promotional benefits are subject to the checkout page.

“Do I have to buy Pro or Max to use more models?”

This is precisely one point worth noting about it: the four plan tiers have the same model access scope; the main difference is quota, not that lower-priced tiers open access to fewer models. Specific available models still depend on the real-time catalog, credential permissions, and client configuration.

Therefore, if you only want to first connect Claude Code or Codex and try it with your own projects, there is no need to buy a higher tier right away.

The meaning of ¥99 is that it gives your first attempt a clear budget. It does not mean you can use it without limits, nor does it guarantee that the quota can meet an entire month's workload, but you do not need to pay ¥699 first for the same model access scope.

Before you have figured out your own usage volume, is it necessary to pay first for future heavy-use needs?

So how long does ¥99 actually last?

This is a good question, but I cannot casually give you an answer such as “how many times you can ask per day.”

Even for “help me fix a bug,” in one situation it may involve changing a dozen lines of code, while in another it may require reading multiple files, calling tools, and making repeated revisions. The input context, output length, model choice, and caching situation are all different, so the consumption naturally differs as well.

Ace meters uncached input, model output, cache writes, and cache reads separately, so 400 Credits cannot be directly converted into a fixed number of prompts.

Rather than listening to someone say, “I cannot even use it all in a month,” it is better to judge with tasks you would actually do:

  • Add tests to a module currently being maintained;
  • Find a cross-file bug;
  • Review a real code change.

First see whether the task can be completed, then see whether the consumption is acceptable. After all, you buy a plan to get the work done, not to buy a quota number that looks very large.

My suggestion is: first purchase the ¥99 basic subscription and try it with your own projects for a period of time. The “trial” here is a paid experience, not a free trial. Focus on whether the integration is smooth, whether the model performance meets expectations, and the actual quota consumption.

If it feels good after using it, you have already started treating it as a daily development tool, and the basic quota is not enough, then consider choosing a ¥299 or higher-tier subscription. Confirm that it is useful first, then increase the budget—is that not more prudent than buying a high-tier plan from the start?

What kind of people are worth putting it on their shortlist?

I would first think of two kinds of people.

The first kind: those who are already using Claude Code or Codex, but want more model choices.

You are already comfortable with the tool and do not want to adapt to a whole new set of operations. What you need is an additional source of models, rather than collecting another new AI editor.

Ace's configuration center provides integration instructions for Claude Code and Codex respectively, including CLI and VS Code. First check the guide corresponding to your current environment, and you can more specifically judge whether it is suitable.

The second kind: those who do not want to maintain multiple subscriptions long term just to occasionally use a certain model.

Perhaps you normally use only one model, and only want to switch to another approach when you encounter a difficult problem. Is it worth subscribing for another month just for this occasional need?

At this point, multi-model plans are at least worth comparing. Their value is not a promise that every call will be the cheapest, but that your choices do not have to be completely constrained by “who you have already paid for this month.” Conversely, if you only use one model over the long term, and your current plan is already sufficient, then there is no need to spend extra money just for “more choices.”

Having more features and being useful to you are two different things.

There are also two rules that I recommend understanding before purchasing

One is the balance.

Ace supports carrying over remaining Credits and extending the validity period by 30 days when you renew before expiration. But if you do not renew when it expires, unused credits will become invalid; purchasing again after expiration will not restore the old balance.

For people whose projects have uneven workloads and who plan to keep using it, this rule is worth considering.

But do not, in turn, buy a new plan that you cannot use up at all just to preserve a small remaining balance. The prerequisite for saving money is that you actually need it.

The other is the billing boundary.

The Coding Plan does not renew automatically, nor does it automatically use the general balance as a backup; meanwhile, it still has normal API concurrency and rate limits. So do not interpret it as “you can run Agents without limits after purchasing it.”

Making these rules clear is more useful than simply saying “use it with confidence.”

So, is it worth buying?

If what you are looking for is the lowest price, unlimited usage, and a plan that works in every scenario, then this article cannot make such a promise to you.

But if your needs are clear:

You want to continue using Claude Code or Codex, want to retain choices across multiple model ecosystems, and also want to use ¥99 first to control your trial budget—then the AceDataCloud Coding Plan is worth a look.

There is no need to rush into choosing the most expensive one.

First confirm the combination of models and clients you want to use, then buy a basic subscription tier and try it with your own real projects. If it feels smooth, the results are satisfactory, and the quota is not enough, then consider a higher tier.

Whether a plan is suitable ultimately does not depend on how lively its introduction page is, but on whether it can fit into your daily development work at a cost you can accept.

If you are interested, you can start with the official plan page below. First look at the prices, then see how your commonly used tools can be integrated. After completing these two steps, it is often more helpful than reading ten recommendation articles.

  1. AceDataCloud Coding Plan Details: Source for the prices, basic quota, model access scope, billing, and renewal rules in this article; for the latest prices and promotional benefits, refer to the order page.
  2. Coding Plan Configuration Center: Provides integration methods, protocol descriptions, and configuration guides for clients such as Claude Code and Codex.