OpenCode Pricing: Plans, Subscription and API Costs (2026)
OpenCode itself is free; the plans you pay for are model access. The OpenCode Go subscription costs $10 a month (Go Plus $40) and gives each model a monthly dollar allowance, with at most 20% of it per 5 hours and 50% per week. OpenCode Zen has no subscription: you pay each model's per-token price plus a card fee on top-ups. Your own API key bills at that provider's rates. Standard Compute is a cost-optimized alternative from $19 a month: it sends each step to the cheapest model that can do it, on one monthly compute budget that stops when used, so it is not unlimited either.
OpenCode lists Standard Compute natively through models.dev, with the base URL https://api.stdcmpt.com/v1. This page covers what OpenCode costs. To choose a model provider, see which API provider to use with OpenCode. Keep running out of Go's windows? See OpenCode Go alternatives.
OpenCode plans and subscription prices compared
| Plan | Price | Limits | Models |
|---|---|---|---|
| OpenCode client | Free, open source | None from OpenCode; the provider you connect sets them. | Any supported provider |
| OpenCode Zen | Pay per token, plus 4.6% + $0.31 per top-up | Your prepaid balance and any budgets you set | About 80, including Claude, GPT and Gemini |
| OpenCode Go | $10 a month | A dollar allowance per model ($15 of Kimi K3 or GLM-5.3, $60 of MiniMax M3, $80 of DeepSeek V4.1 Flash), 20% of it per 5 hours, 50% per week. One member. | 29, mostly open-weight, plus Claude Haiku 5.5, GPT-6 Luna, Grok 4.7 and Gemini 4 Argon |
| OpenCode Go Plus | $40 a month | Same windows, higher allowances ($60 of Kimi K3, $120 of GLM-5.3, $160 of DeepSeek V4.1 Flash). One member. | Same 29 models |
| Your own API key (BYOK) | The provider's per-token rates | The provider's per-minute rate limits and spend caps | That provider's models |
| Standard Compute | From $19 a month with a $20 compute budget | One monthly budget for every model, no 5-hour or weekly windows; requests stop at 100% (HTTP 402), no overage | Claude, GPT, Gemini and open models; the router picks the cheapest one that can do each step |
Go and Go Plus checked October 17, 2026 on OpenCode's Go docs, after OpenCode dropped Grok 4.6, Kimi K2.6 and MiniMax M2.7 from Go on October 16; Zen prices and card fee checked October 13, 2026 on OpenCode's Console docs. Allowances and model lists can change.
OpenCode Zen pricing
Zen is OpenCode's pay-as-you-go model service; its docs now call it OpenCode Console. You add a balance and pay each model's per-token price, which OpenCode says it sells at cost. On top of each top-up comes a card processing fee: 4.6% + $0.31 in the billing docs, so a $20 top-up costs $21.23 (the models page still says 4.4% + $0.30). When the balance drops below $5, Zen reloads $20 unless you change or turn off auto-reload, and at zero paid requests fail with HTTP 402.
| Model | Input | Output | Cached read |
|---|---|---|---|
| Claude Opus 5.5 | $4.00 | $20.00 | $0.20 |
| Claude Sonnet 5.5 | $2.00 | $10.00 | $0.20 |
| Claude Haiku 5.5 | $0.10 | $0.50 | $0.01 |
| GPT 6 Astra | $10.00 | $50.00 | $1.00 |
| GPT 6 Sol | $2.00 | $10.00 | $0.20 |
| GPT 6 Luna | $0.10 | $0.50 | $0.01 |
| Gemini 3.1 Pro | $2.00 | $12.00 | $0.20 |
| Kimi K3 | $3.00 | $15.00 | $0.30 |
| GLM 5.3 | $1.40 | $4.40 | $0.26 |
| DeepSeek V4 Pro | $1.74 | $3.48 | $0.145 |
| DeepSeek V4.1 Flash | $0.30 | $1.20 | $0.006 |
| MiniMax M3 | $0.30 | $1.20 | $0.06 |
Dollars per million tokens, from OpenCode's Zen model list on October 13, 2026. GPT, Gemini and Haiku prices are for the shorter context tier. The full list has about 80 models, including free models offered for a limited time.
OpenCode Zen vs Go
Go suits one developer on open models who rarely hits a 5-hour window. Zen suits teams and anyone who needs Claude or GPT, but you choose the model for every task and the bill follows your usage. Standard Compute sits between them: a fixed monthly plan that includes Claude, GPT and open models, with a router that picks one per request.
| Zen | Go | Go Plus | |
|---|---|---|---|
| Price | Pay per token | $10 a month | $40 a month |
| Models | About 80, incl. Claude, GPT, Gemini | 29, mostly open-weight | 29, mostly open-weight |
| Limits | Your balance and budgets | Per-model monthly dollar limit, 20% per 5 hours, 50% per week | Same windows, higher limits |
| Team | Workspaces, budgets per member, SSO | One member | One member |
Go can fall back to your Zen balance at its limits if you turn that on. Sources: Go docs, budgets, SSO.
What OpenCode Go includes
Go is optional: you do not need a Go subscription to use OpenCode. Its current model list spans multiple families, and some models receive lower effective usage allowances than the base limit. Compare the specific model you want rather than assuming every model includes the same dollar value.
For use in another coding client, check Go's request-identification and session-header requirements. A compatible API format alone does not establish that every client or workload meets a provider's terms.
Estimate your own month
- Choose a small task representative of your repository work and use a clean branch.
- Record task success, elapsed time, tool-call failures and actual provider usage. Repeat before treating one result as typical.
- Estimate how often you run that task and account for longer sessions and parallel agents. Compare the resulting usage with a subscription's actual allowance.
- Confirm what happens at the limit: a stop, an optional purchase, or usage-based charges. Choose spending controls deliberately.
Compare with measured coding-agent usage in the Agent Burn Index
The catch
A fixed monthly price is not unlimited model use. Go has usage windows and model-specific allowances. Standard Compute has an enforced monthly compute budget. Usage-based providers charge for consumption and may offer their own limits and spending controls.
Where Standard Compute fits
Standard Compute is built to cut what each OpenCode task costs: with the model standardcompute, the router sends each step to the cheapest model that can do it, and for every model we vet providers for price, uptime and speed, with automatic fallback when one fails. Opt-in Europe or US Compliance rules keep a key, team or company on providers in that region, and we don't train on your prompts or store them. OpenCode lists it natively, so you connect it with /connect. Standard Compute starts at $19/month with a $20 monthly compute budget. Requests stop when that budget is exhausted until renewal or a plan change; there are no automatic overage charges. Requests are not slowed as the budget is used.
FAQ
Is there an OpenCode subscription?
Yes, two: OpenCode Go at $10 a month and Go Plus at $40 a month. Both give each model a monthly dollar allowance (on Go, $15 of Kimi K3 or GLM-5.3 up to $80 of DeepSeek V4.1 Flash), with at most 20% of it per 5 hours and 50% per week, for one member. The OpenCode client itself needs no subscription.
How much is OpenCode per month?
$0 for the client. Model access is $10 a month on Go, $40 on Go Plus, whatever you spend on Zen or your own API key, or a fixed plan from another provider such as Standard Compute, which starts at $19 a month with a $20 compute budget.
Which OpenCode plan should I get?
Go if you work alone on open models and your usage fits its 5-hour and weekly windows. Zen or your own key if you need Claude or GPT for most tasks and want to pay per token. A routed plan such as Standard Compute if you want frontier and open models on one monthly budget with no 5-hour or weekly windows, at a higher entry price than Go.
What happens when I hit my OpenCode Go limit?
Requests to that model are refused until its 5-hour, weekly or monthly window rolls over. If you turned on the option, Go falls back to your Zen balance and bills per token instead. Other models keep their own allowances.
Is OpenCode free?
The OpenCode client is free and open source. The model provider you connect determines model-access costs. OpenCode Go is optional.
What is the best subscription for OpenCode?
Compare OpenCode Go's native setup and model-specific allowances with other supported providers. The best choice depends on task quality, API compatibility and your actual use; an existing subscription or light usage-based access may be enough.
Is there an unlimited plan for OpenCode?
Do not interpret a flat price as an unlimited allowance. Standard Compute requests stop when the monthly compute budget is exhausted, and Go applies its own usage windows. Check each provider's current terms.
How much does OpenCode Zen cost?
Zen has no subscription. You prepay a balance and pay each model's per-token price, for example $1.40 per million input tokens for GLM 5.3 and $4.00 for Claude Opus 5.5, plus a card processing fee on each top-up (4.6% + $0.31 in OpenCode's billing docs). Workspaces have no seat fees.
Do I need OpenCode Go?
No. OpenCode's documentation describes Go as optional; you can connect other supported providers. Confirm the API protocol and model configuration supported by your installed client.