Home / Guides / OpenCode pricing
OpenCode pricing: free, Go, Go Plus and pay-as-you-go
OpenCode itself is free and open source, with free models included. OpenCode Go is $10 a month, Go Plus $40 and Go Teams $20 per member, all for the same models, about 30, with different limits. Claude Sonnet, Opus and the GPT Sol models are not on Go: for those, buy Console credits at token prices or connect your own provider.
Go's limits are written in dollars per model rather than one number per plan, so the plan table below is followed by the figures for the models people ask about.
Checked 4 November 2026 against OpenCode Go's documentation, Console models and Console billing · By Soren Starck
| Option | Price | What you get | Limit |
|---|---|---|---|
| OpenCode with free models | $0 | The open source agent and a set of free models, no account needed | A daily request limit |
| Go | $10 a month | About 30 mostly open models, one member | $15 to $100 of usage a month, by model |
| Go Plus | $40 a month | The same models, one member | $60 to $240 of usage a month, by model |
| Go Teams | $20 a member a month | The same models, every member of a workspace | $20 to $105 per member, by model |
| Console pay-as-you-go | Token prices, plus 4.6% + $0.31 per top-up | Claude, GPT, Gemini, Grok and open models, no seat fee | Your prepaid credits |
| Your own provider | What the provider charges | API keys, or a ChatGPT Plus or Pro, GitHub Copilot or GitLab Duo plan | The provider’s |
Is OpenCode free?
Yes. The agent is open source under the MIT license, and it runs in the terminal, on the desktop or in an IDE at no cost. It comes with free models you can use without creating an account, up to a daily request limit. The list changes because most are free for a limited time, and the terms differ: Big Pickle, Pulse Alpha Free, Flint 40B Free and MiMo-V2.6-Flash Free may use your data to improve the model, and the NVIDIA Nemotron endpoints say not to send personal or confidential data. Read those terms before pointing one at work code.
What do Go, Go Plus and Go Teams cost?
Go is $10 a month and Go Plus $40, both for a workspace with a single member. Go Teams is $20 per member per month for workspaces with several members, and a member who joins mid-period is charged $20 and gets a full month's allowance. You can cancel at any time and keep the plan until the paid period ends. A refund is possible within 14 days of the charge if you have used almost none of that period's allowance.
Go Plus is not simply four times Go. The price is four times higher, and most models get four times the limit, but GLM-5.3 gets eight times, some like MiniMax M3 get three times, and DeepSeek V4.1 Flash and MiMo-V2.6-Flash get two times. Go Teams adds $5 a month to Go's limit on every model, so a member pays twice Go's price for a little more usage. The extra cost buys a plan that can cover a whole workspace, not much more allowance.
What are the OpenCode Go limits?
Each model has a monthly limit in dollars, measured at the token prices below. A five-hour window can use 20% of it, a week 50% and a month all of it. So two full weeks use up a month, and Go's $15 limit on Kimi K3 means $3 per five hours and $7.50 a week. The last column is OpenCode's own estimate for typical use on Go.
| Model | Input, per 1M | Output, per 1M | Go, a month | Go Plus | Go Teams | Go requests per 5 hours |
|---|---|---|---|---|---|---|
| Kimi K3 | $3 | $15 | $15 | $60 | $20 | 110 |
| Kimi K3.1 Pro | $2 | $10 | $15 | $60 | $20 | 147 |
| Qwen3.8 Max | $2 | $6 | $15 | $60 | $20 | 160 |
| Grok 4.7 (up to 200K tokens) | $2 | $6 | $15 | $60 | $20 | 169 |
| GLM-5.3 | $1.40 | $4.40 | $15 | $120 | $20 | 220 |
| Gemini 4 Argon (half price to 30 Nov) | $2 | $10 | $30 | $120 | $35 | 440 |
| DeepSeek V4 Pro (peak) | $1.32 | $3.96 | $15 | $60 | $20 | 1,050 |
| MiniMax M3 (up to 512K tokens) | $0.30 | $1.20 | $60 | $180 | $65 | 3,200 |
| Claude Haiku 5.5 (up to 100K tokens) | $0.10 | $0.50 | $15 | $60 | $20 | 3,850 |
| GPT 6 Luna (up to 272K tokens) | $0.10 | $0.50 | $15 | $60 | $20 | 4,230 |
| DeepSeek V4.1 Flash (peak) | $0.30 | $1.20 | $100 | $200 | $105 | 43,330 |
These are 11 of Go's models, which number about 30. DeepSeek's off-peak prices are half the peak ones shown here; peak hours are 01:00 to 04:00 and 06:00 to 10:00 UTC, Monday to Friday. Gemini 4 Argon goes back to $4 and $20 after 30 November 2026. The full list is in the Go documentation.
Against token prices, Go is a discount by design: $10 buys between $15 and $100 of usage a month, depending on the model. It only pays off if your usage fits that model's windows. OpenCode says the lower limits sit on models that are new or already discounted, where it has no bulk discount to pass on.
What happens when you hit a Go limit?
By default Go blocks the request until the window resets, and you can keep working on the free models. If you turn on Extra Usage on the Go page and hold Console credits, requests fall back to your pay-as-you-go balance instead. Go Teams starts with Extra Usage on, so a team that does not want overage has to turn it off.
What does Console pay-as-you-go cost?
The Console has no seat fee at any team size. You buy prepaid credits ($10, $50, $100 or another amount) and each request is charged at the model's token price. OpenCode adds a processing fee of 4.6% + $0.31 per purchase and says it charges nothing beyond that, so a $50 top-up costs $52.61. It is the way to run models Go does not include. Prices per million tokens:
| Model | Input | Output | Cached read | Also on Go |
|---|---|---|---|---|
| Claude Fable 5.1 | $10 | $50 | $0.25 | No |
| Claude Opus 5.5 | $4 | $20 | $0.20 | No |
| Claude Sonnet 5.5 | $2 | $10 | $0.20 | No |
| Claude Haiku 5.5 (up to 100K tokens) | $0.10 | $0.50 | $0.01 | Yes |
| GPT 6 Astra (up to 272K tokens) | $10 | $50 | $1 | No |
| GPT 6.1 Sol (up to 272K tokens) | $2 | $10 | $0.10 | No |
| GPT 6 Luna (up to 272K tokens) | $0.10 | $0.50 | $0.01 | Yes |
| Gemini 3.8 Flash | $1.50 | $7.50 | $0.15 | No |
| Grok 4.7 (up to 200K tokens) | $2 | $6 | $0.50 | Yes |
| Kimi K3 | $3 | $15 | $0.30 | Yes |
| DeepSeek V4.1 Flash | $0.30 | $1.20 | $0.006 | Yes |
At zero the balance does not go negative: paid requests fail with HTTP 402. Auto-recharge can top it up from a rule you set, and it switches off if a charge fails, so a long run can still stop. A workspace without a budget has no monthly cap beyond its credits. For what the same Claude models cost elsewhere, see Claude pay-as-you-go pricing.
What happened to OpenCode Zen?
The Zen addresses still work but redirect: opencode.ai/zen opens the Console's models page and the Zen documentation opens the Console models documentation, and neither uses the name any more. OpenCode's home page calls the service OpenCode Models. If you are looking for Zen pricing, the table above is it.
Can you use your own subscription or API key?
Yes, and then OpenCode costs you nothing extra: usage through your own provider is billed by that provider. OpenCode's provider docs list ChatGPT Plus, GitHub Copilot and GitLab Duo as subscriptions that work without setup. Claude Pro and Max are the exception. Plugins for them exist, the same docs say Anthropic explicitly prohibits this, and OpenCode stopped bundling them in version 1.3.0. For Claude models, that leaves an Anthropic API key or Console credits.
Which OpenCode option should you pick?
Start with the free models if you are trying it on code you can share. If you already pay for ChatGPT or Copilot, connect that first, since it adds no cost. Take Go at $10 if open models suit your work and you want a fixed bill; move to Go Plus when one model's window keeps stopping you and that model gets a real increase in the plan. If you need Claude Sonnet, Opus or Fable, or GPT Sol, Go will not help: use Console credits or a provider of your own, and compare Claude's own plans if Claude is most of your usage. Teams with two or more people need Go Teams or Console; larger companies can ask about Enterprise, which has no public price.
Seeing how much you have left
The Console's Billing drawer shows live Go limits, credits and usage, and opencode stats shows tokens and cost for your sessions; how to check OpenCode usage goes through it. Since 2.2.2 the terminal tells you the reset time once you have hit a limit.
SessionWatcher shows your OpenCode Go five-hour, weekly and monthly windows in the Mac menu bar or the Windows taskbar, so you see the limit coming before a task stops, next to the other AI coding tools you use. It reads usage only, not your code or conversations.
SessionWatcherSee your OpenCode Go limits
before a task stops.
Native app for the Mac menu bar and the Windows taskbar. Track Claude, Codex and 20 more tools' usage, costs, and rate limits in real time.
“Fast, simple, and does exactly what it should. Definitely worth it.”
@nicojerome on GitHub
macOS 14+. Free for 7 days, no credit card.
Frequently asked questions
Is OpenCode free?
Yes. OpenCode is open source under the MIT license and free to use, and it comes with a set of free models you can use without creating an account, up to a daily request limit. You pay only if you add a paid plan or connect a paid provider.
How much is OpenCode Go?
Go is $10 a month, Go Plus is $40 a month and Go Teams is $20 per member per month. All three cover the same models, about 30, at the same token prices. They differ in how much usage each model gets, and Go and Go Plus are for one-member workspaces.
What are the OpenCode Go limits?
Each model has a monthly limit written in dollars, from $15 to $100 on Go. Five hours can use 20% of it and a week 50%. On Go that comes to about 110 Kimi K3 requests or 4,230 GPT 6 Luna requests per five hours, by OpenCode’s own estimate of typical usage.
Does OpenCode Go include Claude Opus or GPT Sol?
No. The only Claude model on Go is Claude Haiku 5.5, and the only GPT models are GPT 6 Luna and GPT 5.6 Luna. Claude Sonnet, Opus and Fable and the GPT Sol models need Console credits or your own provider.
What happened to OpenCode Zen?
The Zen pages now redirect to OpenCode Console’s models page, which does not use the name Zen. That page holds the pay-as-you-go model list and prices: prepaid credits, token prices and no seat fee, with a 4.6% + $0.31 fee on each credit purchase.
Can you use a Claude Pro or Max subscription with OpenCode?
OpenCode’s docs say plugins exist for this and that Anthropic explicitly prohibits it, and OpenCode stopped bundling them in version 1.3.0. Its docs list ChatGPT Plus, GitHub Copilot and GitLab Duo as subscriptions that work without setup. For Claude, use an API key or Console credits.
Sources
All read on 4 November 2026:
- OpenCode: free to use, free models, and which subscriptions work.
- OpenCode Go: prices, free model limits, cancelling and refunds.
- Go documentation: the three plans, usage windows, limits and estimated requests per model.
- Console models: pay-as-you-go prices and the data terms of each free model.
- Console billing: credits, the processing fee, auto-recharge and payment methods.
- Console budgets: monthly spend caps.
- OpenCode providers: subscriptions, and the note on Claude Pro and Max.
- OpenCode Enterprise: contact form, no price.