OpenCode Go
Low-cost lane for open coding models
As of 2026-10-07, OpenCode lists Go at $10/month, plus a higher-limit Go Plus at $40/month. Limits are per-model monthly dollar amounts: the 5-hour window is 20% of the monthly limit, the weekly window 50%. Works in OpenCode or any agent via the OpenCode Go provider; one subscriber per workspace. Includes DeepSeek V4.1 Flash, DeepSeek V4 Pro, Kimi K3, GLM-5.3, Qwen3.8 Max, GPT 6 Luna, Grok 4.7 and more.
Updated 2026-10-08
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 / AMEX), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Highlights
Per month
As of 2026-10-07 the official page lists $10/month only, with no $5 first month listed. Cancel any time; top up credit if needed. Go Plus is $40/month.
5-hour window
Each model has a monthly dollar limit: 5 hours 20%, week 50%, month 100%. On Go, DeepSeek V4.1 Flash gets $60 a month and Kimi K3 $15; on Go Plus, $120 and $60.
DeepSeek V4.1 Flash per 5h (estimate)
Official estimate from typical usage (about 410 input, 71,300 cached and 310 output tokens per request); actual usage varies. Go Plus: 52,000.
Different product
Zen is pay-as-you-go; Go is a $10/month subscription. With Use balance on, Go falls back to your Zen balance once its limits are reached.
What Go is
As of 2026-10-07, OpenCode Go is the OpenCode team's (Anomaly) low-cost subscription for a curated set of tested open coding models: Go at $10/month, Go Plus at $40/month. Limits are per-model monthly dollar amounts; the request counts on the official page are estimates from typical usage, so compare the dollar limits, not just the counts.
Why it is trending
The official estimates as of 2026-10-07 are far apart: on Go, Kimi K3 gets about 110 requests per 5 hours and Grok 4.7 about 169, while DeepSeek V4.1 Flash gets about 26,000 and DeepSeek V4 Flash about 13,000. The gap comes from each model's token prices and monthly limit; Flash-class models carry the volume.
Timeline
OpenCode's Go docs updated (page marked Last updated: Oct 6, 2026): Go $10/month, Go Plus $40/month, with monthly dollar limits listed per model.
Checked against the official Go page: it lists $10/month and Go Plus at $40/month, with no $5 first month listed.
End date of the current DeepSeek ZDR agreement per the Go docs (renewed monthly; the Go page FAQ still says 2026-09-30).
Confirmed vs caution
Confirmed
Verifiable in the official docs as of 2026-10-07: Go $10/month and Go Plus $40/month; the per-model rule of 20% per 5 hours, 50% per week and 100% per month; the model list and endpoints; the privacy table. DeepSeek ZDR: the Go docs (updated Oct 6, 2026) say it is renewed monthly and the current agreement is valid through October 31, 2026; the FAQ on opencode.ai/go still says September 30, 2026, so this page follows the newer docs.
Caution
Homepage request counts are estimates, the model list can change and the ZDR date will expire. As of 2026-10-07 the official page lists no $5 first month, so don't take older $5 first-month figures at face value.
How to choose
Want cheap open models
Go, and prefer Flash/MiniMax-class ids inside the cap.
Want one frontier model only
Buying official APIs à la carte may be simpler than Go’s mixed menu.
How to use it
① Sign in to OpenCode Console, subscribe to Go or Go Plus and copy your API key. ② Run /connect in the TUI, pick OpenCode Go and paste the key. ③ Run /models to see what Go offers; in config use opencode-go/<id>, for example opencode-go/kimi-k3. You can also call the endpoints under opencode.ai/zen/go/v1/ directly (chat/completions, messages or responses, depending on the model).
On QCode
OpenCode Go is a third-party subscription from Anomaly and QCode does not resell it. QCode takes a different route: a pay-as-you-go multi-model API key, with the catalogue on /models currently spanning Claude, GPT, DeepSeek, Kimi, GLM and Qwen.
FAQ
Go vs Zen?
No. In the official words, Zen is pay-as-you-go, while Go costs $10/month with generous limits and a curated model lineup.
Really tens of thousands of Flash requests?
They are estimates. The official table as of 2026-10-07 gives DeepSeek V4.1 Flash on Go about 26,000 requests per 5 hours and 130,000 per month, assuming about 410 input, 71,300 cached and 310 output tokens per request. The real cap is the model's monthly dollar limit ($60 on Go).
Team seats?
Docs: only one member per workspace can subscribe to Go or Go Plus.
Includes DeepSeek V4 Flash?
Yes. As of 2026-10-07 the official list includes DeepSeek V4 Flash, plus DeepSeek V4.1 Flash and DeepSeek V4 Pro; run /models in the OpenCode TUI to see the current list.
After limit?
Optional Zen balance, else free models.
Any agent?
Yes. The docs say Go works with any agent and list clients validated besides OpenCode, such as Claude Code, Codex and Hermes, without guaranteeing they keep working. Your client should send typical coding-agent traffic, identify itself with its own user agent and send a stable session ID per conversation.
Sources
OpenCode's official Go page (opencode.ai/go) and Go docs (opencode.ai/docs/go/, marked updated Oct 6, 2026), fetched 2026-10-07; the docs win on numbers.
Compare dollar limits, not request estimates
Prefer per-token billing with one key for claude-sonnet-5-5, gpt-6.1-sol, deepseek-v4.1-flash, kimi-k3 and more? Try QCode; see /models for each model's price.
Related
Go + DeepSeek
Flash/Pro ids and ZDR.
Go Free
Free tier vs paid Go.
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.
Checked against OpenCode's official Go page and docs on 2026-10-07; the official pages take precedence. Not affiliated with Anomaly/OpenCode; limits change. QCode model availability is whatever /models shows.