Buildful Pass: a $30 coding plan for Kimi K3, DeepSeek, GLM and MiniMax

The Buildful Pass is a $30 a month coding plan you add to Pro or Team. It gives you $80 of usage a month on four open-weight coding models: Kimi K3, DeepSeek V4.1 Flash, GLM 5.3 Flash and MiniMax M3.1 Flash. Pick one of them for a task, and it runs on the Pass instead of your prepaid credit until the $80 is spent.

Smart Mode is still the default on every plan, and for most tasks it's the right choice: it picks the model for you, and your plan's tasks cover it. The Pass is for when you want to choose the open-weight model yourself, task by task, without topping up credit first.

What is the Buildful Pass?

The Buildful Pass is $80 of model usage a month for $30, on four models:

All four are open-weight models, which most developers call open source: their makers publish the weights, so many companies can serve them, which keeps their prices low. DeepSeek V4.1 Flash is the model that replaced DeepSeek V4 Flash in September 2026. You can add the Pass on Pro ($10 a month) or Team ($40 a month). It isn't on the Free plan. More models may join the Pass later.

In the task box, the four models carry a Pass tag while your Pass has allowance left.

How does the allowance work?

You get $80 of usage each billing period, and each task you run on a Pass model is charged against it at that model's usage cost, the same way prepaid credit is charged. A few rules decide the rest:

• The Pass renews with your plan: the two start, renew and end on the same day.
• If you add it mid-month, it's prorated. You pay for the rest of the period and get the same share of the $80.
• Unused allowance doesn't roll over, so each period starts at $80 again.
• When the $80 runs out, a task that's already running finishes, and the next task on a Pass model draws from your prepaid credit.
• Your own key comes first. If your team has added a Kimi API key, Kimi K3 bills to that key instead of the Pass.
• Only a team owner can add, cancel or resume the Pass, from Billing. Cancelling takes effect at the end of the period.

Models you pick yourself, the Pass ones included, don't use your plan's Smart Mode tasks. Those stay for Smart Mode.

When should you use the Pass, Smart Mode, credit or your own key?

For most of your backlog, stay on Smart Mode. It picks an open-weight model for each task, and your plan covers every one: 1,000 tasks a month on Pro, 3,000 on Team.

Switch to a Pass model when you know which one you want. Kimi K3 is the obvious case: with a 1M token context, you can point it at a large repo and let it read. The Pass is also how you try the four side by side without topping up credit for each run.

Frontier models like Claude Opus 5.5, GPT-6 Astra and Gemini Argon run on prepaid credit. Top up from $5, and each task is charged at the model's usage cost.

If you already pay OpenAI, Azure OpenAI, Anthropic, xAI or Kimi, add your API key in settings and those models bill to your account instead.

Every model you can pick on Buildful

On Pro and Team, the task box lists Smart Mode and Quick Mode, then 20 models from 10 companies. Each one runs in the same cloud sandbox with the same tools: your checks, a browser test for UI changes, and a pull request at the end.

Smart Mode and Quick Mode are the only options on the Free plan. The open-weight models page has more on the open-weight side, and Buildful pricing has the plans.

How does the Pass compare with other coding plans?

Three other coding plans sell access to the same kind of models.

On price alone, OpenCode Go is the lowest way in, and the Pass needs a Buildful plan underneath it, so the smallest total is $40 a month.

The difference is what you get at the end. Those plans give you a model to drive from your own agent, on your own machine. The Pass runs the model inside Buildful: a cloud sandbox, your checks, a browser test for UI changes, and a pull request on your repo. It works as a background coding agent, with nothing running on your laptop.

Frequently asked questions

Which models does the Buildful Pass include?

Kimi K3, DeepSeek V4.1 Flash, GLM 5.3 Flash and MiniMax M3.1 Flash. More may join later.

Does the Pass use my Smart Mode tasks?

No. Tasks on Pass models draw from the Pass's $80. Your plan's tasks stay for Smart Mode and Quick Mode.

What happens when the $80 runs out?

The task that's running finishes. After that, the four models draw from your prepaid credit until the Pass renews with your plan.

Does unused allowance roll over?

No. Each billing period starts with $80.

Can I use my own Kimi API key instead?

Yes. If your team has added a Kimi key, Kimi K3 bills to your key, and the Pass covers the other three models.

Is Kimi K3 good for coding?

For large codebases, yes. Kimi K3 is Moonshot AI's open-weight model with a 1M token context, so it can hold a big repo in view while it works. On Buildful it runs as a full coding agent: your checks, a browser test for UI changes, and a pull request at the end.

How much does Kimi K3 cost on Buildful?

Kimi K3 pricing on Buildful depends on how you run it. With the Pass, it draws from your $80 a month until that's spent. After that it's charged to prepaid credit at its usage cost, or to your own Kimi API key if you've added one. If you'd rather run Kimi on your own machine, Kimi Code's plans start at $19 a month.

Which plans can add the Buildful Pass?

Pro and Team. It's an add-on to a paid plan, so the Free plan can't add it.

What's the best LLM for coding?

On Buildful, for most tasks, Smart Mode: it picks the model per task, and your plan covers it. For the hardest work, pick a frontier model like Claude Opus 5.5 or Fable 5.1. For a large repo you want read end to end, try Kimi K3 on the Pass.

To add the Pass, open Billing in Buildful. If you don't have an account yet, start building on Buildful.