OpenCode and Vana API: one address, $0.10 setup credit

§ One address for OpenCode and your other tools

Can OpenCode and your other tools really point at the same place? With Vana API, the answer is yes, and the detail that makes it practical is simple: one address, https://api.vanaapi.com, three protocols. You do not need to hunt for a different endpoint for every client in your workflow. The same Vana API address answers API requests in the OpenAI format, the Anthropic Messages format and the Gemini native format, and each tool talks to it in the way that tool already knows how to talk.

OpenCode is one of the clients that fits the OpenAI format, so it sits alongside Cursor, Cline, Roo Code, Cherry Studio, any OpenAI SDK and cURL. That means you can move between tools without rebuilding your setup each time you change your mind about which editor to use. Vana API is not a separate product per tool. It is one service, one address, and a small set of clearly defined protocols.

One clarification keeps the picture tidy: https://api.vanaapi.com answers API requests only. Keys, balance and the usage log live in the dashboard on the main site, https://vanaapi.com. So when you set up OpenCode, you create a key in the dashboard, copy it into the client, and give the client the Vana API address. There is no second billing page to manage and no separate account for each tool you like.

§ Which client speaks which protocol?

Which protocol belongs to which client? Vana API keeps that mapping narrow and explicit, which is what makes the setup short. Pick the client you already use, and use the protocol that client speaks.

• Claude Code and Claude Desktop, version 2.1.285 or later, speak the Anthropic Messages protocol (/v1/messages) themselves. Those clients do not reach the other formats through an SDK, cURL or a script.
• Gemini CLI uses the Gemini native protocol (/v1beta/models/...). It is not OpenAI-compatible, so it stays on its own route to the same Vana API address.
• GPT, Grok, DeepSeek, GLM, Kimi, Qwen, MiniMax and MiMo models use the OpenAI format (/v1/chat/completions) from Cursor, Cline, Roo Code, OpenCode, Cherry Studio, any OpenAI SDK or cURL. GPT models also answer on /v1/responses, which is what Codex uses.

For most people reading this, OpenCode is the line that matters. It connects over the OpenAI format, it accepts a base address, a key and a model id, and it can reach the GPT, Grok, DeepSeek, GLM, Kimi, Qwen, MiniMax and MiMo models listed in the Vana API price list. You are not translating requests or wrapping anything. The client sends the format it sends, and Vana API routes it.

§ Setting up OpenCode with minimal steps

What does the setup actually involve? Say you run OpenCode every day and also keep a second tool open for longer sessions. You would create one key in the Vana API dashboard, enter https://api.vanaapi.com as the base address in each client, paste the key, and choose the model id you want. The same key and the same address work for both tools, because both speak the OpenAI format.

There is no monthly subscription to start and no seat to buy. Vana API runs on a prepaid balance, and calls are charged per real token. You top up the balance when you want to, and you can watch what happened in the usage log in the dashboard. Every call is listed there, so if you want to know what a given session cost, the numbers are written down for you.

Signing up also puts $0.10 of credit on the account. That is a small amount rather than a headline offer, but it is enough to complete a test request through OpenCode and confirm that the address, the key and the model id all line up before you commit real spend.

§ What the models cost through Vana API

Is the pricing easy to follow? Prices are listed per million tokens as input and output, and the same list covers every client that speaks the OpenAI format, including OpenCode. Here is the Vana API price list, with the model ids you would enter in the client.

• GPT-5.6 Sol (model id gpt-5.6-sol): $0.3472 / $1.81, against an official list price of $4 / $20.
• GPT-6 Astra (model id gpt-6-astra): $0.6944 / $3.54, against an official list price of $10 / $50.
• GPT-6.1 Sol (model id gpt-6.1-sol): $0.1389 / $0.8333, against an official list price of $2 / $10.
• Grok 4.7 (model id grok-4.7): $0.4861 / $1.53.
• DeepSeek V4 Flash (model id deepseek-v4-flash): $0.2778 / $1.04.
• DeepSeek V4 Pro (model id deepseek-v4-pro): $1.11 / $3.33.
• GLM-5.3 (model id glm-5.3): $0.5556 / $1.67.
• Kimi K3 (model id kimi-k3): $1.04 / $5.21.
• Qwen3.7 Plus (model id qwen3.7-plus): $0.1278 / $0.4167.
• Qwen3.8 Max (model id qwen3.8-max): $0.6944 / $2.08.
• MiniMax M3 (model id MiniMax-M3): $0.1389 / $0.5556.
• MiMo V2.6 Pro (model id MiMo-V2.6-Pro): $0.2083 / $0.3472.

For example, say you run a long OpenCode session on GPT-6.1 Sol every day and you want a rough sense of the input side. The listed input rate is $0.1389 per million tokens, and the output rate is $0.8333 per million tokens. Both figures come straight from the Vana API price list, so you can compute a working estimate for your own traffic rather than guessing from a headline number.

Because the price list is the same for every OpenAI-format client, switching from OpenCode to Cursor or Cline does not change your rates. You are not pricing each tool separately, and you are not buying a bundle that forces one model. You choose the model id that suits the task and pay the listed rate.

§ Balance, plans, referrals and refunds

How does the money work day to day? You top up a prepaid balance in the Vana API dashboard, and each call draws down that balance by the real tokens it consumed. There is no monthly fee attached to the balance itself, and the usage log in the dashboard records every call so you can line spending up with the work you actually did.

If your usage is steady and you would rather plan ahead, Vana API also offers a monthly plan for GPT models: GPT up to $4,000 / 30 days costs $103.61 for 30 days, at most $1,200 a week, and it covers GPT models only. That option exists alongside the pay-as-you-go balance rather than replacing it, so a light week still works the same way as any other Vana API call.

Inviting other people works on the same dashboard. You receive 14% of what people you invite pay, and 5% one level further. Referral earnings can be withdrawn from $13.89. If you would rather not think about it, the referral programme simply sits there while your normal usage continues.

What if you top up and change your mind? Unused balance can be refunded within 3 days of the top-up, minus what was used and the payment processor's fee. So the practical sequence is short: create a key in the dashboard, point OpenCode at https://api.vanaapi.com, pick a model id from the price list, and let the prepaid balance and the usage log handle the accounting.