OpenCode Go Free
Free models: daily limit, lineup and when Go pays off
As of 2026-11-09, OpenCode's free models have a daily request limit, but the official FAQ gives no number; the only daily figure in the docs is 100 requests per day for LongCat 2.5 Free (on 2026-10-15 the same FAQ said 200 requests a day). There is no plan called Go Free: Go is a paid $10/month subscription, Go Plus is $40/month and Go Teams is $20 per member per month. All free models are offered for a limited time and the lineup changes, and free-tier chat models work only inside OpenCode. When the daily limit is not enough, you can subscribe to Go, add Console credit or connect a per-token API (see below).
Updated 2026-11-13
Prices change: on QCode, Claude, GPT and Chinese models are billed per token — live rates on /models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Highlights
Free / Zen free
Official FAQ (checked 2026-11-09): free models have a daily request limit, but no number is given; the Console docs mark them all as limited-time and list names such as Big Pickle, GLM-5.3-Flash Free and DeepSeek V4.1 Flash Free.
Go
As of 2026-11-09: Go $10/month, Go Plus $40/month, Go Teams $20 per member per month; each model has a monthly dollar limit, with 20% per 5 hours, 50% per week and 100% per month.
LongCat 2.5 Free (limited time)
As of 2026-11-12 this is the only daily number in OpenCode's docs: LongCat 2.5 Free is limited to 100 requests per day, is hosted in China and needs Global regions in the workspace privacy settings. On the Go list, LongCat 2.5 Free, Beam Free and Iceman Contributor Free show no monthly cap (limited time); Crouton Alpha Free was deprecated on 2026-11-09. DeepSeek V4.1 Flash is pay-as-you-go on Console ($0.30 per 1M input) and counts against limits on Go.
After Go's limits
Without Extra Usage you can keep using the free models; with it, Go draws on your Console credit instead of blocking requests.
What “Go Free” usually means
As of 2026-11-09, OpenCode's official pages list no plan called Go Free: the Go docs name three monthly plans, Go ($10/month), Go Plus ($40/month) and Go Teams ($20 per member per month), while the free option is OpenCode's free models, which have a daily request limit and are offered for a limited time. People searching "opencode go free" usually want a $0 coding agent, so separate the two first, then decide when an upgrade pays off.
Limits on OpenCode's free models
As of 2026-11-12 there are four layers of limits. Quota: the Go page FAQ says free models have a daily request limit but gives no number (on 2026-10-15 it said 200 requests a day); the docs state a number only for LongCat 2.5 Free, 100 requests per day. Time: the Console docs list Big Pickle, Iceman Free, GLM-5.3-Flash Free, DeepSeek V4.1 Flash Free, MiMo-V2.6-Flash Free and the other free models as available for a limited time, and Pulse Alpha Free was deprecated on 2026-11-06; Crouton Alpha Free was deprecated on 2026-11-09. Data: it differs per model; for Big Pickle, Flint 40B Free and MiMo-V2.6-Flash Free, collected data may be used to improve the model, Iceman Free retains collected data and uses it to improve the model, Inkling 2 Small Free retains data and may use it for training, GLM-5.3-Flash Free and DeepSeek V4.1 Flash Free follow a zero-retention policy and do not train on your data, and NVIDIA's free Nemotron endpoints are for trial use only, with no personal or confidential data. Region: for Inkling 2 Small Free and Beam Free the docs list countries and regions where the model is not available, Russia among them. Go subscriptions follow a different rule: per-model monthly dollar limits enforced across rolling windows (20% per 5 hours, 50% per week, 100% per month).
Timeline
Go docs updated (page marked Last updated: Oct 6, 2026): Go $10/month, Go Plus $40/month; after Go's limits you can keep using the free models.
Checked against the official Go page FAQ: the free-model quota is 200 requests a day.
Checked against the Go page FAQ, the Go docs and the Console docs: the FAQ no longer gives a daily number (on 2026-10-15 it still said 200 requests a day, when the Console free-model table listed Big Pickle, MiMo-V2.6-Flash Free, Step 5 Preview Free and others); LongCat 2.5 Free is the only model with a stated daily limit, 100 requests; Go Teams has been added; the Go docs now live at opencode.ai/docs/console/go/.
Confirmed vs caution
Confirmed
Verifiable as of 2026-11-12: free models have a daily request limit (the Go page FAQ gives no number) and are all offered for a limited time; LongCat 2.5 Free is limited to 100 requests per day; free-tier chat models require OpenCode, so other clients need paid models or Go; Go $10/month, Go Plus $40/month and Go Teams $20 per member per month, with per-model monthly dollar limits; after Go's limits you can continue on free models or turn on Extra Usage to draw on your Console credit; LongCat 2.5 Free, Beam Free and Iceman Contributor Free are also on the Go list with no monthly cap shown (limited time); Crouton Alpha Free was deprecated on 2026-11-09.
Caution
Two common misconceptions: not every free model id offers a 1M context window, and the free lineup changes with promotions, so a free model in your config may disappear (Pulse Alpha Free was deprecated on 2026-11-06). The 200-requests-a-day figure that many pages still repeat was in the Go page FAQ on 2026-10-15 but could not be found there on 2026-11-09; other precise numbers circulating online mostly come from individual testing, not official commitments.
How to choose
Stay on the free models
Exploration, learning and low-stakes work: fine if you can live with a daily request limit OpenCode does not state as a number, a changing lineup and code that is not sensitive (some free models may use free-period data to improve the model).
Subscribe to Go ($10/month)
A daily-driver agent with a curated lineup and documented per-model monthly limits, usable in other agents too; heavy users should look at Go Plus ($40/month), and teams at Go Teams ($20 per member per month). Go does not include Claude Opus or Claude Sonnet; for those you need per-token billing.
How to use it
① Run /models in OpenCode and pick Big Pickle or a model with Free in its name; free-tier chat models work only in OpenCode, so other clients need paid models or Go. ② When the daily limit is not enough, subscribe to Go ($10/month), Go Plus ($40/month) or Go Teams ($20 per member per month) and use /connect with OpenCode Go; or add credits on Console and pay as you go, or connect your own provider's API key (OpenCode ships a provider and model catalog from models.dev, and for most providers you run /connect first and then pick a model). ③ To hide free models from the list, set OPENCODE_DISABLE_FREE_MODELS=1 on the server process.
When the daily limit runs out: a per-token API
When the free models' daily limit runs out, or Go is not an option for you, connect a per-token API in OpenCode and keep working. QCode is a multi-model API key, separate from OpenCode: billed per token, per-model rates are listed on /models, and there is no 4-hour window. One key reaches Claude, GPT and leading Chinese models, for example deepseek-v4.1-flash, glm-5.3, qwen3.8-flash, kimi-k3 and claude-sonnet-5-5. In OpenCode, Claude uses the anthropic provider with baseURL https://api.qcode.cc/claude/v1 and GPT uses https://api.qcode.cc/openai/v1; Chinese models need their own @ai-sdk/openai-compatible provider with that same OpenAI baseURL. To fill in the key, run /connect, pick Anthropic (or OpenAI), choose Manually enter API Key, paste the QCode key and press Enter, then /exit and reopen OpenCode; OpenCode 1 keeps the credentials in ~/.local/share/opencode/auth.json and OpenCode 2 keeps them in the server's SQLite database. Follow the OpenCode page at docs.qcode.cc/en/docs/ide/opencode. Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 days; if you already have a Personal Plan, add a booster pack when you need more quota. Details: qcode.cc/pay-as-you-go. Payment options include credit cards, Apple Pay, Google Pay, Alipay, WeChat Pay and crypto.
FAQ
What are the limits on OpenCode's free models?
Free models have a daily request limit, but as of 2026-11-09 OpenCode's pages give no single number: the Go page FAQ names none (it said 200 requests a day on 2026-10-15), and the docs state a number only for LongCat 2.5 Free, 100 requests per day. Besides the quota there are three more limits: every free model is offered for a limited time and the lineup changes; some free models may use data to improve the model; and a few are not available in some countries and regions.
Is Go Free a plan name?
No. As of 2026-11-09, official Go has three monthly plans: Go ($10/month), Go Plus ($40/month) and Go Teams ($20 per member per month). "Go Free" is mostly a search mashup of Go and the free models.
What does “free usage limit reached” mean, and how do I fix it?
It is the daily request cap on the free models, not a fault and not an account problem. As of 2026-11-12 the official repository carries several quota-related messages, so identify yours first: ① "Free usage limit reached. Connect a provider or subscribe to OpenCode Go to keep going, or switch models." — the free models' quota for today is used up (the code path is FreeUsageLimitError) and the dialog shows when your free messages reset and links to the Go subscription page, https://opencode.ai/go. Fixes: wait for the reset, add your own provider with /connect, switch models, or subscribe to Go. ② "Free usage exceeded" — the compact label for the same free quota in the session view, and the “Free limit reached” dialog that follows gives the reset time and the subscribe entry. ③ "OpenCode Go … usage limit reached. Switch to a free model, or turn on Extra Usage on the Go page in OpenCode Console to keep going." — this one is a Go subscription limit (GoUsageLimitError), not the free quota, and Console names the window in its response; the fixes are to switch to a free model or turn on Extra Usage on the Go page. "Workspace usage limit reached" is a different line: that is a workspace budget cap, so wait for the budget to reset or ask an owner or admin to raise it. The usual routes are still the ones above: subscribe to Go ($10/month), Go Plus ($40/month) or Go Teams ($20 per member per month) and work within per-model monthly dollar limits; add credits on Console and pay as you go; or use /connect to add your own provider's API key. If you want another per-token route, see “When the daily limit runs out” above.
Are “quota”, “$5” and the free models the same thing in OpenCode?
No. The free models cost nothing and have only a daily request limit. There is no $5 plan: Go starts at $10/month (Go Plus $40/month, Go Teams $20 per member per month), and its allowances are monthly dollar amounts per model, counted over 5-hour, weekly and monthly windows; the only $5 in the Go docs is that Go Teams limits are $5/month higher than Go on every model. Console prepaid credit is separate again: the top-up presets are $10, $50 or $100, or choose Other and enter an amount, with a 4.6% + $0.31 processing fee added on top, and the balance is spent on usage. After a Go limit you can still use the free models; as of 2026-11-12 the Go list also includes LongCat 2.5 Free, Beam Free and Iceman Contributor Free, with no monthly cap shown (limited time). Console budgets are a separate switch as well: free models do not count toward them, and neither does Go usage within the plan; Go Extra Usage paid from credit does count.
Do free models train on my code?
It depends on the model. Per the Console docs as of 2026-11-12: for Big Pickle, Flint 40B Free and MiMo-V2.6-Flash Free, data collected during the free period may be used to improve the model; Iceman Free retains that data and uses it to improve the model; Inkling 2 Small Free retains data and may use it for training; the providers of GLM-5.3-Flash Free and DeepSeek V4.1 Flash Free follow a zero-retention policy and do not train on your data; Beam Free retains data for 30 days and does not train on it. NVIDIA's free Nemotron endpoints are for trial use only, so don't submit personal or confidential data. Keep sensitive code away from free models that use data to improve the model.
How do I pay for paid models, and what if I can't?
OpenCode's billing docs (as of 2026-11-09) list card, Link, Alipay and WeChat Pay among the payment methods, added under Settings › Billing in Console, and Go subscriptions use the same methods; the docs do not say which countries or cards are accepted, so the checkout page decides. If you want another per-token route, use QCode: it accepts credit cards (Visa / Mastercard), Apple Pay, Google Pay, Alipay, WeChat Pay and crypto, and it is fully self-service. Billed per token; per-model rates are listed on /models. Set it up in OpenCode with the OpenCode page at docs.qcode.cc.
Sources
The FAQ on OpenCode's official Go page (opencode.ai/go, now redirecting to opencode.ai/console/go), the Go docs (opencode.ai/docs/go/, now redirecting to opencode.ai/docs/console/go/), the free-model table and privacy notes in the Console models docs (opencode.ai/docs/zen/, now redirecting to opencode.ai/docs/console/models/), the Models and Providers docs and the billing docs (opencode.ai/docs/console/billing/), fetched 2026-11-09.
Outgrowing the free models?
QCode bills per token, with one key for deepseek-v4.1-flash, glm-5.3, kimi-k3 and more; see /models for prices.
Related
OpenCode Go
Paid lane.
Cline 400 with a custom Anthropic base URL: the fix
4.1.22 started sending the fallbacks option and custom endpoints returned 400; 4.1.23 (2026-10-07) fixes it. Plus the QCode Anthropic-provider setup.
Flash 0731
Why Flash is hot.
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
Checked against OpenCode's official Go page, Go docs, Console models docs and billing docs on 2026-11-09; the official pages take precedence. Not affiliated with OpenCode; free quotas and the lineup can change without notice. QCode model availability is whatever /models shows.