OpenCode Go Pricing, Limits, Models, and Setup

Updated August 13, 2026.

OpenCode Go is a subscription for selected coding models inside OpenCode and other compatible agents. The first month costs $5, then $10 per month. The plan provides a usage budget of up to $12 per 5 hours, $30 per week, and $60 per month; the actual number of requests depends on the selected model.

In brief: Go is a good fit for a developer who regularly uses DeepSeek V4, GLM, Kimi, Qwen, MiniMax, MiMo, or Grok 4.5 and is willing to route tasks accordingly. It is not unlimited and it is not the same as OpenCode Zen: Go is a fixed subscription with time windows, while Zen is a pay-as-you-go balance.

This article is intended for developers and small teams. Prices, models, and limits change; below is the status of the official Go page and documentation as of August 13, 2026.

Contents

• How OpenCode Go works
• Pricing and limits
• Which models are included
• Why the number of requests differs
• OpenCode Go vs. Zen: what’s the difference
• How to activate the subscription
• Who benefits from the subscription
• Limitations and privacy
• FAQ
• How AI Dawn helps you build an AI development stack
• Bottom line

How OpenCode Go works

OpenCode remains a free open-source coding agent: you can connect your own API keys from any supported provider. Go is an optional provider that gives you a single API key and access to tested model/provider combinations.

After payment, the user selects OpenCode Go via /connect, pastes the key, and sees the models through /models. The subscription can be used not only in OpenCode, but also in compatible agents through OpenAI- or Anthropic-compatible endpoints.

Only one participant can subscribe in a single workspace. For teams, this means Go should not automatically be treated as a full per-seat enterprise license: you should check the member management and billing model before rolling it out broadly.

Pricing and limits

Parameter	Value
First month	$5
Subsequent months	$10/month
5-hour limit	$12 usage value
Weekly limit	$30 usage value
Monthly limit	$60 usage value
Cancellation	at any time

The limits are expressed as the dollar cost of inference, not as the same number of messages. If a model is expensive, the window will run out faster. If it is cheaper, you’ll get more requests. All three windows apply at the same time: an unused monthly balance does not let you exceed the 5-hour limit.

When Go is exhausted, you can continue on free models. If OpenCode Zen has a pay-as-you-go balance, the Use balance setting switches requests to it after the limit is reached instead of blocking them.

Which models are included

As of August 13, the official list includes:

• Grok 4.5;
• GLM-5.2 and GLM-5.1;
• Kimi K3, Kimi K2.7 Code, and Kimi K2.6;
• MiMo-V2.5 and MiMo-V2.5-Pro;
• MiniMax M3 and MiniMax M2.7;
• Qwen3.7 Max, Qwen3.7 Plus, and Qwen3.6 Plus;
• DeepSeek V4 Pro and DeepSeek V4 Flash;
• Hy3.

The marketing calls Go a plan for open coding models, but Grok 4.5 is not an open-weight model. It is more accurate to think of Go as a curated set of available coding models, not a guarantee of the same license or the ability to download the weights.

The list may change after testing, updates, and agreements with providers. Do not build a mission-critical process around a single model without a fallback.

Why the number of requests differs

The documentation publishes request estimates based on a typical Go pattern. These are not guaranteed quotas: context length, output, reasoning, and cache hits all affect usage.

Model	Estimated requests per 5 hours	Per week	Per month
Grok 4.5	120	300	600
GLM-5.2	880	2,150	4,300
Qwen3.7 Max	950	2,390	4,770
DeepSeek V4 Pro	3,450	8,550	17,150
Qwen3.7 Plus	4,300	10,800	21,600
DeepSeek V4 Flash	31,650	79,050	158,150

The difference between Grok 4.5 and DeepSeek V4 Flash is more than two orders of magnitude because the usage limit is shared by cost. That makes routing the main way to get value: fast models handle search, formatting, and simple edits, while expensive ones handle architecture and complex debugging.

OpenCode Go vs. Zen: What’s the Difference?

Criterion	OpenCode Go	OpenCode Zen
Billing model	subscription	pay as you go
Starting point	$5 for the first month, then $10	add funds to the balance, usually $20
Limits	$12/$30/$60 windows	set balance and customizable monthly limits
Models	limited Go list	broader curated catalog
After the limit	stop/free or use balance	charged against balance
Best for	regular work with included models	occasional requests, premium models, and precise spending control

A practical setup is to use Go as the base plan and Zen as a fallback for models outside the list or for peak usage. Zen auto top-up should be configured carefully: by default, when the balance falls below $5, the system may add $20, which can lead to charges beyond the logical monthly limit.

How to Activate the Subscription

1. Install or update OpenCode.
2. Launch the TUI and enter /connect.
3. Select OpenCode Go.
4. Open the sign-in page, log in, and subscribe.
5. Copy the API key and paste it into the TUI.
6. Run /models and select a model.
7. In the console, check the usage windows and, if needed, configure use balance.

Do not commit the key to opencode.json or the repository. Store it in a secret manager or a secured environment variable. For teams, issue separate credentials where supported, and review the spending log regularly.

Who Benefits from the Subscription

Go is a good fit if you spend more than $10 per month on models from the list, are willing to use a low-cost route for most tasks, and do not need one specific closed frontier model.

The subscription may not be a good fit if:

• you need a Claude/OpenAI flagship model that is not available in Go;
• your work is occasional and Zen PAYG is cheaper;
• your company policy requires a specific data region and contract terms;
• the monthly $60 usage value runs out long before the end of the period;
• your process critically depends on an unchanged model ID.

To estimate it, take a week of logs: model, input/output tokens, cache, task type, and share of accepted changes. Compare the actual spend with $10 and check how many tasks can safely be moved to Flash/Plus.

Limitations and Privacy

OpenCode says Go is aimed at international users, and the models are hosted in the U.S., EU, and Singapore. Providers follow a zero-retention policy and do not use data for training — this is the service’s statement, which should be checked against your contract and your organization’s requirements.

Zero retention does not eliminate local logs, agent telemetry, terminal history, or data in connected tools. Do not send production secrets just because an API has a privacy statement. Mask credentials, limit context files, and confirm commands that change infrastructure.

FAQ

How much does OpenCode Go cost?

$5 for the first month, then $10 per month. Terms may change, so check the official checkout before paying.

Is OpenCode Go unlimited?

No. Usage value windows apply: $12 for 5 hours, $30 per week, and $60 per month. The number of requests depends on the model and tokens.

Does it include DeepSeek V4?

Yes, the current list includes DeepSeek V4 Pro and V4 Flash. According to OpenCode’s estimate, Flash delivers many more requests within the same usage window.

Can you use Go outside OpenCode?

Yes. The official page allows you to use the plan with any compatible agent through the provided endpoints.

How is Go different from free OpenCode?

OpenCode itself is free to use and supports your own keys. Go is a paid provider with curated models and an included usage budget.

How AI Sunrise Helps Build a Practical AI Development Stack

AI Sunrise can audit your development process and data, match models to task classes, set up a corporate AI workspace, RAG, and integrations, and implement limits, logging, and acceptance criteria. If an external gateway does not meet requirements, the team can evaluate on-premise LLMs and a controlled internal route.

The first step is to choose one process, establish a baseline, define data sources, constraints, and the acceptance criterion. Discuss the project.

Bottom line

OpenCode Go is a rare plan where a $10 subscription can translate into up to $60 in usage value per month, but only within the concurrent limits of $12 per 5 hours and $30 per week. Its value depends on routing: DeepSeek V4 Flash and Plus models deliver a large volume, while Grok 4.5 burns through the window much faster.

Before paying, check the current model list, privacy requirements, and the weekly profile of your tasks. If the Go models fit your needs, the subscription can replace several separate API keys; if you need a broad catalog or irregular usage, Zen PAYG remains the more flexible option.