Claude Fable 5.1 pricing and usage log with Vana API

§ One address, three protocols, and a short path to a Claude Fable 5.1 session

Vana API is built around a simple idea: one address, three protocols, minimal setup. The request address is https://api.vanaapi.com, and it answers API requests. The dashboard, where you create keys, check your balance and read the usage log, lives separately at https://vanaapi.com. Keeping those two roles distinct makes the first steps straightforward, because the endpoint you send requests to and the place you manage your account are not competing for your attention.

For a Claude Fable 5.1 session, the relevant protocol is Anthropic Messages, served at /v1/messages. Claude models are reachable only from official Claude Code or Claude Desktop, version 2.1.285 or later, because those applications speak the Anthropic Messages protocol themselves. That is the entire client requirement for this model, and it also means there is no second integration to assemble before you begin.

The other two protocols exist on the same address for other model families. Gemini models use the Gemini native protocol at /v1beta/models/..., and they are used from Gemini CLI. GPT, Grok, DeepSeek, GLM, Kimi, Qwen, MiniMax and MiMo models use the OpenAI format at /v1/chat/completions. One address therefore covers three protocol styles, while each model still follows the path its own client expects.

• Request address: https://api.vanaapi.com
• Dashboard, keys, balance and usage log: https://vanaapi.com
• Claude models: Anthropic Messages at /v1/messages, from official Claude Code or Claude Desktop, version 2.1.285 or later
• Gemini models: Gemini native protocol at /v1beta/models/..., from Gemini CLI
• OpenAI-format models: /v1/chat/completions, with GPT models also answering on /v1/responses

Before your first session, create a key in the dashboard and add funds to your prepaid balance. Billing is pay-as-you-go: you are charged per real token, and there is no monthly fee. A new account also receives a gift on sign-up of $0.10 of credit, which is enough to see a first request and a first usage log entry rather than guessing what the flow looks like.

§ What the usage log tells you after a Claude Fable 5.1 session

Once your session is finished, the usage log in the Vana API dashboard is the place to review what happened. Every call is listed there. That means the record is not a monthly summary you have to reconcile with your own notes, but a per-call view of the requests your key made while you were working.

Read the log as a set of decisions rather than as a wall of numbers. Start by identifying the calls that belong to the session you just ran. Then check the model identifier attached to each entry. For Claude Fable 5.1 the model id is claude-fable-5-1, and seeing that id confirms that the request was handled by the model you intended to use rather than by a neighbouring option in the model list.

Next, look at the token counts for each call and the resulting charge against your prepaid balance. Because Vana API charges per real token, the log is where a session's cost becomes concrete. You can see which individual requests carried the weight of the session, and whether a long exchange or a short one accounts for most of the spend. The balance shown in the same dashboard tells you what remains.

A practical reading routine looks like this. Open https://vanaapi.com, select the usage log, and scan the newest entries first. Note the model id, the input and output token counts, and the charge for each call. Add the charges in your head as a rough sense of the session total, then compare that with the balance change. If the two do not line up, the log entry is the place to look for an additional call that you may not have registered, such as a retry or a follow-up turn.

For example, say you run this every day and want to check one day's work before you top up again. You would open the usage log, find the entries for that day's Claude Fable 5.1 session, confirm that each shows claude-fable-5-1, and sum the per-call charges. That example is not a forecast; it is simply the review step that turns token usage into a number you can budget against.

§ How the Claude Fable 5.1 price translates into a session total

The price list gives a clear per-million-token basis for Claude Fable 5.1. At Vana API, the input price is $3.33 and the output price is $16.74 per million tokens. The official list price is $10 / $50. Comparing the two lines is useful because it shows the shape of the saving without rounding anything away: the input figure is $3.33 rather than $10, and the output figure is $16.74 rather than $50.

To turn that into a session figure, take the real token counts from the usage log and apply the two rates. For example, say a session used 250,000 input tokens and 10,000 output tokens. The input portion works out from $3.33 per million tokens, and the output portion works out from $16.74 per million tokens. The usage log then shows the charge as it was actually recorded, which is the figure to trust for accounting.

Output tokens carry the higher rate in this model, so a reading of the log benefits from a simple question: how much of the session was output? If a large share of the tokens came back as generated text, the output side of the rate matters more to the total than the input side. The log's per-call breakdown lets you see that pattern across individual requests instead of only at the end of the month.

The same care applies to the other Claude entries you may meet while comparing options. Claude Haiku 4.5 is $0.3472 / $1.60 at Vana API, Claude Opus 5.5 is $1.46 / $6.94, and Claude Sonnet 5.5 is $1.18 / $5.76. Each one carries its own model id in the log, so you can confirm which price applies to which entry rather than assuming that all Claude traffic is billed at a single rate.

§ Deciding between prepaid balance and a Claude monthly plan

The usage log is also the evidence you need when choosing how to pay. If your logged sessions are irregular, or you want to stop after a specific piece of work, a prepaid balance fits naturally. You add what you expect to use, you are charged per real token as calls arrive, and there is no monthly fee to renew. You can see the balance and the per-call charges in the same dashboard, so the decision to top up again stays grounded in recorded usage.

If your logged usage is steady and predictable, a monthly plan can make the arithmetic simpler. Two plans cover Claude models only. The first is Claude up to $1,000 / 30 days: $64.72 for 30 days, at most $320 a week. The second is Claude up to $2,400 / 30 days: $153.61 for 30 days, at most $750 a week. Both are Claude-only, so neither covers the OpenAI-format models or the Gemini native protocol.

The choice follows directly from the log. If your recent sessions sit comfortably inside the first plan's weekly ceiling of $320, the $64.72 plan gives you a fixed frame for 30 days of Claude work. If your logged usage repeatedly approaches that ceiling and stays within the second plan's $750 weekly limit, the $153.61 plan is the closer fit for a larger, sustained workload. If usage arrives in bursts and then stops, the prepaid balance avoids paying for a month you did not need.

Refunds add a useful piece of flexibility to the prepaid route. Unused balance can be refunded within 3 days of the top-up, minus what was used and the payment processor's fee. That window is short enough to keep in mind and long enough to cover a top-up you made slightly ahead of a project that has not yet started. Reading the usage log tells you exactly how much of the balance has been consumed, which is the starting point for any refund calculation.

§ A simple checklist before you top up again

Use the log to answer four questions, in order. Which model ids appear in the session? Which calls account for most of the tokens? What did the session cost against the prepaid balance? And is the pattern steady enough to justify a monthly plan, or irregular enough that pay-as-you-go balance is the better fit?

Three details deserve a second look. Confirm that Claude Fable 5.1 entries show claude-fable-5-1, since the price depends on the model id. Check that the per-call charges add up to the balance movement you expected. And if you are comparing plans, remember the ceilings are weekly: at most $320 a week on the $64.72 plan and at most $750 a week on the $153.61 plan, each for 30 days and each limited to Claude models.

One address, three protocols, minimal setup: the endpoint is https://api.vanaapi.com, the dashboard is https://vanaapi.com, and the usage log records every call. After a Claude Fable 5.1 session, that log gives you the model, the tokens and the charge for each request, and the Claude Fable 5.1 rate of $3.33 / $16.74 per million tokens gives those numbers a price you can check. With the gift on sign-up of $0.10 of credit, referral earnings of 14% of what people you invite pay and 5% one level further, withdrawable from $13.89, the practical next step is simply to open the dashboard and read the entries your session produced.