Claude Sonnet 5.5 Pricing on Vana API: $1.18 input per million tokens

Money mattersEN2026-11-01

§ One address, three protocols, minimal setup

Most multi-model API decisions begin with plumbing. You collect base URLs, key formats, protocol variants, and a separate client for every model family. Vana API takes a different route: one address, three protocols, minimal setup. The single endpoint is api.vanaapi.com, and that address answers API requests only. Keys, balance, and the usage log live in the dashboard on the main site, vanaapi.com, so the working surface stays separate from the account surface and each one stays easy to find.

Three protocols run on that one address. Claude models use the Anthropic Messages protocol at /v1/messages. Gemini models use the Gemini native protocol at /v1beta/models/..., and they are not OpenAI-compatible. GPT, Grok, DeepSeek, GLM, Kimi, Qwen, MiniMax, and MiMo models use the OpenAI format at /v1/chat/completions. GPT models also answer on /v1/responses, which is what Codex uses.

Because the protocol decides the client, you pick the model family first and then point the matching tool at the same address. There is no second base URL to remember, no routing file to write, and no separate account to fund for each family. That is the whole idea behind one address, three protocols: the transport is settled by the protocol your tool already speaks, and Vana API keeps the destination the same in every case.

§ Getting Claude Sonnet 5.5 onto Vana API

Claude Sonnet 5.5 is the model in focus here, and it has one clear rule. It runs only from official Claude Code or Claude Desktop, version 2.1.285 or later, because those tools speak Anthropic Messages themselves. No other client is a supported route for it. If you already use one of those two tools, moving to Vana API is a small step rather than a rebuild.

The setup sequence is short. Create a key in the dashboard on vanaapi.com. Add funds to the prepaid balance. In official Claude Code or Claude Desktop, point the Anthropic-compatible base address at api.vanaapi.com and select the model id claude-sonnet-5-5. That is the configuration work in full: one key, one address, one model id. There is no extra subscription to activate, no separate billing relationship to establish, and nothing to install beyond the client you already run.

Version matters here, and it is worth checking before you top up. Official Claude Code or Claude Desktop 2.1.285 or later is the requirement, so a quick version check now avoids a failed first call later. Everything after that is standard: the same address for every protocol, and a dashboard that shows what each request actually cost.

§ What Claude Sonnet 5.5 costs on Vana API

Here is the number that matters for Claude Sonnet 5.5. On Vana API, the price per million tokens is $1.18 for input and $5.76 for output. The official list price for the same model is $2 / $10. The model id on Vana API is claude-sonnet-5-5, so the pricing entry and the configuration entry use the same identifier.

The gap is straightforward arithmetic. For example, say you process one million input tokens and one million output tokens in a month. On Vana API that is $1.18 + $5.76 = $6.94. At the official list price of $2 / $10, the same volume would be $2 + $10 = $12. The difference is $5.06 for that example month, and it comes from the two published prices rather than from a promotion, a credit, or a quota game.

Because billing is charged per real token, the figure scales with actual use rather than with a seat or a tier you may never touch. Output tokens cost more than input tokens on both price sheets, so a workload that asks for long answers will cost more than one that sends long prompts and receives short replies. Reading the two columns side by side, $1.18 against $2 for input and $5.76 against $10 for output, is enough to plan a budget before you commit to a top-up amount.

§ Prepaid balance, monthly plans, and the usage log

Vana API uses a prepaid balance, charged per real token, with no monthly fee. Every call is listed in the usage log in the dashboard, which means you can check a line item against the token counts you expect instead of guessing at the end of a billing cycle. Keys are created in the dashboard as well, so access control and spending review sit in the same place you already use to top up.

Monthly plans are optional and specific about what they cover. The Claude plan covering up to $1,000 / 30 days costs $64.72 for 30 days, with at most $320 a week, and it covers Claude models only. The larger Claude plan covers up to $2,400 / 30 days for $153.61 for 30 days, with at most $750 a week, also Claude models only. For GPT work, the GPT plan covers up to $4,000 / 30 days for $103.61 for 30 days, with at most $1,200 a week, and it covers GPT models only.

If your usage is steady and you would rather have a known ceiling, a monthly plan gives you one. If usage is occasional or hard to forecast, the prepaid balance keeps the commitment at zero, because there is no monthly fee and you pay only for the tokens that pass through. Sign-up credit of $0.10 is a small start, and it lets you place a first call and open the usage log before you decide on anything larger.

§ Referrals and refunds, in plain numbers

Inviting others is simple arithmetic too. You receive 14% of what the people you invite pay, and 5% one level further. Earnings can be withdrawn from $13.89, so referral credit becomes withdrawable once it reaches that threshold. Referral earnings are separate from the token bill, and they are drawn from what invited accounts actually pay rather than from a fixed allowance.

Refunds are straightforward as well. Unused balance can be refunded within 3 days of the top-up, minus what was used and the payment processor's fee. In practice that makes an early top-up low-risk: if you conclude that Vana API is not the right fit for your workflow, the unused portion can come back inside the window, with the stated deductions applied.

§ The checklist to follow once

Here is the sequence to follow, in order, and then you are running.

Open the dashboard on vanaapi.com and create a key.
Top up the prepaid balance, or pick a monthly plan if your usage is predictable enough to justify one.
Keep api.vanaapi.com as the single API address. It answers API requests only, and the dashboard stays on vanaapi.com.
For Claude Sonnet 5.5, open official Claude Code or Claude Desktop, version 2.1.285 or later, and use the model id claude-sonnet-5-5.
For Gemini models, use Gemini CLI over the Gemini native protocol. They are not OpenAI-compatible.
For OpenAI-format models, use the OpenAI format at /v1/chat/completions. GPT models also answer on /v1/responses, which is what Codex uses.
Review the usage log after the first calls and check the per-token charges against the price of $1.18 input and $5.76 output per million tokens for Claude Sonnet 5.5.
If the balance goes unused, request a refund within 3 days of the top-up.

One address, three protocols, minimal setup. Vana API keeps the endpoint fixed, lets the protocol choose the client, and charges for the tokens that actually pass through, with the full record waiting in the dashboard. For Claude Sonnet 5.5 specifically, the numbers are clear enough to decide on today: $1.18 per million input tokens and $5.76 per million output tokens on Vana API, against an official list price of $2 / $10, with a prepaid balance and no monthly fee when you would rather stay flexible.