Unlimited plans vs metered pricing for security work

A flat monthly price cannot pay for unlimited compute, so a flat plan has to limit something else. Usually that is speed, through fair-use thresholds, a stamina reserve or a queue. This post compares four ways of selling model access: Verboo's monthly plans, Venice.ai's monthly plans with a metered API, Drael's four-week blocks, and NotHoly's prepaid per-token credit. The question is what each one limits, and what that means for security work that arrives in bursts and needs long context.

The facts come from each provider's own pages and terms, collected on 9 to 11 November 2026, and each one is linked where it is cited. Where a provider does not publish a figure, this post says so instead of estimating one. Prices are list prices in US dollars. They change, so check the source before you buy.

#How each service sets the price

Service	Cheapest paid entry	Top of the range	What renews or expires
Verboo	US$ 20 a month, Lite	US$ 269 a month, Ultra, sold as unlimited tokens	Monthly subscription
Venice.ai	US$ 18 a month, Pro, with US$ 1 of API credit	US$ 200 a month, Max, with US$ 225 of API credit	Plans renew automatically; monthly credit expires
Drael	US$ 20 a week, Pro	US$ 1,790 a year, Max at full price	Blocks do not renew on their own; no API on the weekly plan
NotHoly	US$ 5 top-up	Top-ups up to US$ 1,250, 10% off	No subscription; credit does not expire

Venice's plan prices are on its plans page. Drael's plans are sold by the week, the four-week block or the year, and its home page lists each price. A Drael month is 28 days, so a year holds 13 four-week payments. At list prices, 13 blocks of Pro come to US$ 910 a year. Our Drael pricing explainer lists every figure.

#What "unlimited" means in the terms

Verboo's home page headline is "Tokens ilimitados. Preço fixo." (Unlimited tokens. Fixed price.) Clause 8 of its Terms of Use, version 2.4 dated 30 September 2026, says unlimited "não significa capacidade computacional ilimitada, disponibilidade ininterrupta, velocidade garantida" (does not mean unlimited compute, uninterrupted availability or guaranteed speed). Clause 9 says heavy token use alone "não constitui violação de Uso Justo" (is not by itself a fair-use violation), and fair use is judged against internal thresholds the company does not publish.

In the terms, the token count is uncapped. The compute behind it is not.

#Where the limit moves on a flat plan

Service	What limits throughput	Where it is stated
Verboo	Stamina reserve, queuing, prioritization and rate limits	Home page; Terms of Use, clauses 8.1 and 20
Venice.ai	Not published in the sources we collected	Not applicable
Drael	Four answers in flight per account, across all keys; per-minute rate not published	Limits documentation
NotHoly	Per-key rate limits that you set in the console	Console

These are ordinary controls, and each provider states them in its own documents. Verboo's clause 20 lists queuing, rate limits, reduced concurrency and reduced priority among the measures it may apply. Drael's limits page says a plan's priority "buys your place in a line, never a faster answer", and a fifth concurrent request gets a 429 error that is not retried for you. On a fixed price, these are where the real limit sits.

#What a metered bill gives you instead

NotHoly bills per token from prepaid credit. You top up US$ 5, 10, 50, 500 (5% off) or 1,250 (10% off) by card or crypto. The credit does not expire, so a month without work costs nothing. Rate limits are per key, and you set them in the console, so throughput is a number you plan around rather than a reserve that refills on someone else's schedule. There is no stamina reserve and no subscription queue. Per-token rates are on pricing, and each model's base and context window are on models.

The API accepts OpenAI and Anthropic formats at api.notholy.com/v1. You send your own max_tokens and reasoning setting, and the gateway trims max_tokens only when your prepaid balance cannot cover the worst case. Drael's chat documentation says max_tokens is dropped on a turn that reasons, so a call through a key has no output ceiling you set. Our Drael API cost post works through the numbers.

Venice works another way. Its plans carry a monthly credit that expires. Pro does not carry unused credit over, and Pro Plus and Max keep it for two and three months. Its API sits on top of the monthly fee and is always metered, so a heavy month pays for both the plan and the tokens. Its API rates are on Venice API pricing, and our Venice AI vs NotHoly post reads the plans in full.

#A worked example

Take a code review that runs 10 million input tokens and 2 million output tokens in one week. On a flat unlimited plan the bill is the same whether that week runs fast or slow, and the speed depends on the provider's reserve and queue. On a metered plan the same review is billed for the tokens it uses, at the rate for its model, and throughput is the per-key limit you set. One trades a fixed bill for a variable wait. The other trades a predictable wait for a variable bill. Which one is better depends on whether the deadline or the budget is the harder constraint.

#Where each one fits

Verboo fits steady, interactive coding where a fixed monthly bill is worth more than guaranteed speed. Venice fits personal use of a large catalog that includes chat, image, video and music, with a metered API on top. Drael fits one person who is active in every four-week block and prefers a block price. NotHoly fits security work that arrives in bursts, where you want a cost you can work out per task and rate limits you control. It sells no subscription and no fixed monthly bill, so if you want one of those, the other three are where to look. Our Verboo vs NotHoly post covers Verboo's terms in more detail.

#Frequently asked questions

#Does unlimited mean no limits?

No. It means no fixed token quota. The compute behind the tokens is still limited, and each provider manages it with fair use, a reserve, a queue or rate limits. Verboo's clause 8 says as much.

#Is metered billing more expensive?

It depends on volume. Light or bursty use usually costs less metered, because you pay only for what you send and an idle month costs nothing. Heavy, steady use can cost less on a flat plan, if you accept the speed limits that come with it.

#Does NotHoly throttle?

It sets per-key rate limits that you choose in the console, and it meters tokens, not speed. There is no stamina reserve and no subscription queue.

#Is Drael's API cheap for light use?

Not on its published plans. The API needs a four-week or annual plan, so the smallest purchase is US$ 70 per four weeks, and the weekly plan has no API at all. A metered provider with no minimum bills only the tokens you send.

#Verdict

A flat price fixes the bill, not the speed. Verboo's terms say unlimited does not mean guaranteed speed, Venice's monthly credit expires, and Drael's API starts at US$ 70 per four weeks. Metered billing moves the variable back to a token count you can see, with throughput you set yourself. Pick the variable you can live with, and price your real workload on pricing before you commit.