Codex on Vana API: one address, $0.10 credit, 5-minute setup

§ One address, three protocols, minimal setup

Vana API is built around a simple shape: one address, three protocols, minimal setup. The address is https://api.vanaapi.com, and it answers API requests only. Keys, balance, and the usage log live in the dashboard on https://vanaapi.com. That split keeps the setup small. You create a key in the dashboard, point your client at the API address, and choose the protocol your client already speaks.

Three protocols sit behind that single address. Anthropic-style traffic goes to /v1/messages. Gemini-native traffic goes to /v1beta/models/.... OpenAI-format traffic goes to /v1/chat/completions, and GPT models also answer on /v1/responses, which is what Codex uses. You do not need a separate endpoint per vendor, and you do not need a second account for each family of models. One address covers the routing decision for you.

For a five-minute quick start, the order of operations matters less than the boundaries. Set up the key, then the address, then the protocol and model identifier. Vana API bills against a prepaid balance charged per real token, with no monthly fee on the pay-as-you-go side, so the feedback loop is immediate: make a call, open the usage log, and see what it cost.

New accounts receive $0.10 of credit on sign-up. That is enough to send a small request and confirm the path end to end before you top up.

§ Codex setup questions, answered

The most common setup question about Codex on Vana API is which address to use. Use https://api.vanaapi.com. The main site, https://vanaapi.com, is where you create keys, check your balance, and read the usage log, so the API address and the dashboard address do different jobs and should not be swapped.

The second question is which endpoint Codex should call. Codex uses the Responses path, /v1/responses, and GPT models answer there. The same GPT models also answer on /v1/chat/completions, which is convenient when you want to move a request between clients without changing the model.

The third question is which model identifier to enter. From the current list: gpt-5.6-sol, gpt-6-astra, and gpt-6.1-sol. These are the model ids you use on Vana API, and they are the ids you reference on the OpenAI-format path as well.

The fourth question is how billing behaves. There is no monthly fee on pay-as-you-go usage. Vana API charges per real token against a prepaid balance, and every call is listed in the usage log in the dashboard. If a request fails, or a client sends a malformed payload, the log tells you what happened before you top up again.

For example, say one request on GPT-5.6 Sol used 1,000,000 input tokens and 1,000,000 output tokens. At Vana API rates of $0.3472 / $1.81 per million tokens, that example request would cost $0.3472 + $1.81 = $2.1572. Real requests are usually smaller, and the usage log shows the actual figure for each call.

§ Matching the right client to each model

Because the address is shared, the only real compatibility decision is which client you are using and which protocol it speaks. Vana API keeps that mapping narrow and explicit.

• Claude models work only from Claude Code or Claude Desktop, version 2.1.285 or later, because those clients speak the Anthropic Messages protocol (/v1/messages) themselves.
• Those models do not work from any other client, any SDK, cURL, or a script.
• Gemini models work only from Gemini CLI, over the Gemini native protocol (/v1beta/models/...). They are not OpenAI-compatible.
• GPT, Grok, DeepSeek, GLM, Kimi, Qwen, MiniMax, and MiMo models use the OpenAI format (/v1/chat/completions), and work 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.

That makes a mixed setup ordinary. You can keep a Claude workflow inside Claude Code or Claude Desktop version 2.1.285 or later, run Gemini work in Gemini CLI, and use Codex or an OpenAI-format editor for GPT and the other OpenAI-format families, all against the same address and the same key.

For instance, Claude Opus 5.5 (claude-opus-5-5) and Claude Haiku 4.5 (claude-haiku-4-5-20251001) are reached through the Messages protocol, while Gemini 2.5 Flash (gemini-2.5-flash) and Gemini 3.1 Pro (gemini-3.1-pro) are reached through the Gemini native protocol. Both sit behind https://api.vanaapi.com, so switching families does not mean changing addresses.

§ Prepaid balance, real-token billing, and the usage log

Vana API runs on a prepaid balance. You top up, calls draw down the balance as they happen, and there is no monthly fee. Every call is listed in the usage log in the dashboard, so the balance and the log tell the same story: what you spent and on what.

Referral credit is part of the same model. You earn 14% of what people you invite pay, and 5% one level further. Earnings can be withdrawn from $13.89.

Refunds are straightforward within a defined window. Unused balance can be refunded within 3 days of the top-up, minus what was used and the payment processor's fee. That is a short window, so it is worth reading the usage log right after a top-up rather than weeks later.

For teams that prefer a fixed allowance, monthly plans sit alongside the prepaid balance rather than replacing it.

§ Monthly plans and per-token pricing

Three monthly plans are available, and each one is scoped to a single model family.

• Claude up to $1,000 / 30 days: $64.72 for 30 days, at most $320 a week, Claude models only.
• Claude up to $2,400 / 30 days: $153.61 for 30 days, at most $750 a week, Claude models only.
• GPT up to $4,000 / 30 days: $103.61 for 30 days, at most $1,200 a week, GPT models only.

Per-token pricing on Vana API is where the pay-as-you-go side becomes concrete. Prices are per million tokens, input / output: Claude Fable 5.1 (claude-fable-5-1) at $3.33 / $16.74, Claude Haiku 4.5 (claude-haiku-4-5-20251001) at $0.3472 / $1.60, Claude Opus 5.5 (claude-opus-5-5) at $1.46 / $6.94, Claude Sonnet 5.5 (claude-sonnet-5-5) at $1.18 / $5.76, GPT-5.6 Sol (gpt-5.6-sol) at $0.3472 / $1.81, GPT-6 Astra (gpt-6-astra) at $0.6944 / $3.54, and GPT-6.1 Sol (gpt-6.1-sol) at $0.1389 / $0.8333.

The Gemini and OpenAI-format families are priced the same way, per million tokens: Gemini 2.5 Flash (gemini-2.5-flash) at $0.2083 / $1.60, Gemini 3.1 Pro (gemini-3.1-pro) at $1.39 / $7.92, Gemini 3.8 Flash (gemini-3.8-flash) at $1.04 / $4.65, Gemini 4 Argon (gemini-4-argon) at $1.46 / $6.88, Grok 4.7 (grok-4.7) at $0.4861 / $1.53, DeepSeek V4 Flash (deepseek-v4-flash) at $0.2778 / $1.04, DeepSeek V4 Pro (deepseek-v4-pro) at $1.11 / $3.33, GLM-5.3 (glm-5.3) at $0.5556 / $1.67, Kimi K3 (kimi-k3) at $1.04 / $5.21, Qwen3.7 Plus (qwen3.7-plus) at $0.1278 / $0.4167, Qwen3.8 Max (qwen3.8-max) at $0.6944 / $2.08, MiniMax M3 (minimax-M3) at $0.1389 / $0.5556, and MiMo V2.6 Pro (MiMo-V2.6-Pro) at $0.2083 / $0.3472.

§ Your five-minute Vana API checklist

Here is the short path through setup, in the order that avoids rework.

• Open https://vanaapi.com and create a key in the dashboard, where keys, balance, and the usage log all live.
• Point Codex at https://api.vanaapi.com, use the /v1/responses endpoint, and choose a GPT model id such as gpt-6.1-sol or gpt-5.6-sol.
• If you use other clients as well, match each one to its protocol: Claude Code or Claude Desktop version 2.1.285 or later for Claude models, Gemini CLI for Gemini models, and the OpenAI-format clients for the OpenAI-format families.
• Send one small request, then open the usage log in the dashboard to confirm the charge against your prepaid balance.
• Top up once the pattern is clear, and remember that unused balance can be refunded within 3 days of the top-up, minus what was used and the payment processor's fee.

One address, three protocols, minimal setup. Vana API keeps the moving parts small: a single API address, a key you create in the dashboard, and a clear rule about which client speaks to which model.