OpenCode Go: Plans, Models and DeepSeek V4.1 Flash Pricing
OpenCode Go is the optional subscription inside OpenCode: three monthly tiers, 34 models, and usage limits counted in dollars. Here are the prices, how the allowances work, the DeepSeek V4.1 Flash rate card and the setup.
OpenCode Go in short
- It is a subscription inside OpenCode, not a separate app: "OpenCode Go gives you reliable access to popular open coding models, with three monthly plans."
- Go is $10/month, Go Plus $40/month and Go Teams $20/member/month. Go and Go Plus serve single-member workspaces; Go Teams bills each member separately.
- Allowances are monthly dollar amounts, capped at 20% per 5 hours and 50% per week. Every response reports what the request consumed in the opencode.allowance field.
- DeepSeek V4.1 Flash on those plans: $0.15/$0.60 per 1M tokens off-peak and $0.30/$1.20 at peak, with monthly allowances of $100 (Go), $200 (Go Plus) and $105 per member (Go Teams).
OpenCode Go — OpenCode docs
Sourceopencode.ai2026-11-16
Models & Pricing
Sourceapi-docs.deepseek.com2026-11-16
Plans, prices, allowances, model names and client compatibility on this page were read from the two sources above on 2026-11-16. Where a number is not on those pages, it is not on this page either.
What OpenCode Go is
The docs define it in one sentence: "OpenCode Go gives you reliable access to popular open coding models, with three monthly plans." You subscribe, you get an API key, and Go behaves like any other provider in OpenCode's /connect picker.
It is also optional in the strict sense: the documentation says it "is completely optional and you don't need it to use OpenCode". Free and bring-your-own-key routes still exist, so Go is a way to buy a fixed monthly allowance across a curated model list, not a paywall in front of the agent.
The three plans and what they cost
Plan names, prices and the one-line description of what each tier buys, taken from the docs.
| Plan | Price | What you get |
|---|---|---|
| Go | $10/month | Lower-cost access to the models listed on this page. |
| Go Plus | $40/month | Higher usage limits across the same models. |
| Go Teams | $20/member/month | Every member gets their own limits — $5/month higher than Go on every model. |
Go and Go Plus are for single-member workspaces such as the Personal workspace a new console creates; both can upgrade to Go Teams at any time, and the unused part of the current month converts into allowance. Go Teams covers every member of a multi-member workspace at $20 each, and a member who joins mid-month pays $20 for a full month of allowance. Service accounts are not covered. Pay with a card or with workspace credit; usage past the allowance runs as Extra Usage, and workspaces on invoiced billing cannot pay for Go with credit.
The full model list
The models the docs list under Go, in their order, with the two DeepSeek entries highlighted. The page adds that "the list of models may change as we test and add new ones": Beam Free, Iceman Contributor Free and LongCat 2.5 Free are marked time-limited, and Beam Free plus the two Muse Spark Contributor models are region-limited.
- Grok 4.7
- GLM-5.3-Flash
- GLM-5.3
- GLM-5.2
- GPT 6 Luna
- GPT 5.6 Luna
- Claude Haiku 5.5
- Gemini 4 Argon
- Kimi K3.1 Pro
- Kimi K3
- Kimi K2.7 Code
- LongCat-2.0
- Step 5 Preview
- Beam Free
- Iceman Contributor Free
- LongCat 2.5 Free
- MiMo-V2.6-Flash
- MiMo-V2.6-Pro
- MiMo-V2.5
- MiMo-V2.5-Pro
- MiniMax M3.1 Flash
- MiniMax M3
- Muse Spark 1.3 Contributor
- Muse Spark 1.2 Contributor
- Qwen4 Max
- Qwen3.8 Max
- Qwen3.8 Flash
- Qwen3.7 Plus
- DeepSeek V4.1 Flash
- DeepSeek V4 Pro
- Hy4
- Hy4 preview
- Hy3
How the limits actually work
There is no request counter to watch. Allowances are defined as monthly dollar amounts, and each model draws from its own line.
- 1Every model has three ceilings over that allowance: 5 hours — 20% of the monthly limit, weekly — 50%, and monthly — 100%.
- 2Each response reports what the request consumed in the opencode.allowance field, so you can see the draw per call instead of guessing.
Token prices are identical across Go, Go Plus and Go Teams — the plans differ in the allowance, not in the rate. Anything beyond the allowance bills as Extra Usage.
DeepSeek V4.1 Flash on OpenCode Go
Per 1M tokens. DeepSeek V4.1 Flash sits in the Go model list; these are its rates and its monthly allowance inside each plan.
| Tier | Off-Peak | Peak |
|---|---|---|
| Cached read | $0.003 | $0.006 |
| Input | $0.15 | $0.30 |
| Output | $0.60 | $1.20 |
Monthly allowance for DeepSeek V4.1 Flash
| Plan | Monthly allowance |
|---|---|
| Go | $100 |
| Go Plus | $200 |
| Go Teams | $105 |
Peak hours are 01:00-04:00 and 06:00-10:00 UTC, Monday through Friday; all other hours, including weekends, are Off-Peak. DeepSeek's own pricing page adds that Chinese public holidays are excluded from peak. DeepSeek V4 Pro runs on the same plans at $0.66/$1.98 per 1M tokens off-peak and $1.32/$3.96 at peak, with monthly allowances of $15 (Go), $60 (Go Plus) and $20 per member (Go Teams).
Subscribe and connect in four steps
Everything happens between the OpenCode Console and the OpenCode TUI — no config file editing.
- 1
Subscribe in the OpenCode Console
Sign in to the OpenCode Console, choose Go, Go Plus or Go Teams, and fill in the billing details. The subscription attaches to the workspace you subscribe in.
- 2
Copy the API key
The console issues an API key for the subscription. Copy it — it is the credential the agent authenticates with, exactly like any other provider key.
- 3
Run /connect and pick OpenCode Go
In the TUI, type /connect, select OpenCode Go from the provider list and paste the key when prompted.
- 4
Run /models and choose a model
Then type /models and pick from the Go list. DeepSeek V4.1 Flash and DeepSeek V4 Pro are both there, provided Regions is set to Global.
/connectOpenCode Go/modelsThe one privacy setting DeepSeek's models need
DeepSeek V4.1 Flash and DeepSeek V4 Pro only work when Regions is set to Global under Settings › Privacy in the console. The default allowed regions are Europe & United States, and endpoints that train on request data are blocked by default — with those defaults the two DeepSeek models cannot be selected.
Where else a Go key works
The docs say Go is designed for OpenCode and for coding agents that produce similar requests, and they publish a short compatibility table.
Clients listed as verified:
- Hermes (v0.21.6+)
- Claude Code
- Codex
- ZCode
- Pi
- jcode (v0.81.6+)
- Kilo Code CLI
Known problematic clients
DeepSeek Harness leads the "known problematic clients" list: session information is missing on some model routes and present on others. That makes it officially known-problematic rather than supported — check the docs before routing production work through it.
