Sulat.com / Writing on AIThursday, November 19, 2026
Agent Engineering/18 November 2026

Capy’s new plans include up to 300,000 machine minutes and 2,500 review credits a month, from a US$1 first week

Capy swapped its single credit balance for machine minutes, review credits and model usage. Lite is US$20 a month with 30,000 minutes; the post's 2,500 reviews are Max's low-effort count.

The Capy logo as a 3D object between server racks, a stopwatch and review cards

At a glance

  • Capy's new Lite plan costs US$20 a month and includes 30,000 machine minutes, 250 review credits and US$20 of model usage, according to its pricing page.
  • The post's 300,000 minutes and 2,500 review credits belong to the US$200 Max plan; the US$1 is the price of Lite's first week.
  • At the default Medium review effort Max's 2,500 credits cover 500 reviews; the post's 2,500 is the low-effort count.
  • On the default 4 CPU, 16 GB machine, Lite's 30,000 minutes last 125 hours.
  • Capy's post lists three unlimited free models; the pricing page doesn't mention them, and the models page still listed two at per-token prices on 18 November.

Capy has swapped its single credit balance for three separate monthly allowances: machine minutes, review credits and model usage. The new Lite plan costs US$20 a month and includes 30,000 machine minutes, 250 review credits and US$20 of model usage, according to Capy’s pricing page. Capy announced the plans on X on 17 November.

The post’s headline figures are the top of the range. The 300,000 minutes and 2,500 review credits belong to the US$200 Max plan, and the US$1 is the price of Lite’s first week.

What each plan includes

On 14 November, when I compared Capy’s US$20 plan with Amp’s, Lite was one US$20 balance that paid for tokens, machine time and reviews alike. Each plan now lists three pools instead, plus the biggest machine it can start. Unused minutes and credits don’t roll over, and every plan has unlimited members with no per-seat charge.

PlanPrice a monthMachine minutesReview creditsModel usageLargest machine
FreeUS$0None100NoneYour own computers
LiteUS$2030,000250US$204 CPU, 16 GB
ProUS$100150,0001,250US$1008 CPU, 32 GB
MaxUS$200300,0002,500US$20016 CPU, 64 GB
Monthly allowances from Capy’s pricing page.
Capy Lite, Pro and Max compared: machine minutes, review credits, model usage, biggest machine
Each paid plan sets minutes, review credits, model usage and the largest machine.

Pro and Max also add cloud phones and mobile builds. Phones bill from credits while awake, at US$5.40 an hour for an iPhone or iPad and US$3.60 for Android.

What a machine minute buys

A machine minute is one minute on a 1 CPU, 4 GB machine. Bigger machines burn minutes faster, one a minute for every 4 GB of RAM. The default Large machine has 4 CPUs and 16 GB, so it uses 4 minutes a minute. On Large, Lite’s 30,000 minutes last 7,500 minutes, or 125 hours, Pro’s last 625 hours and Max’s last 1,250. On Max’s biggest machine, 16 CPUs and 64 GB, 300,000 minutes last 312.5 hours. A sleeping machine uses none, and each time one goes to sleep it adds a two-minute wind-down.

Capy Lite's 30,000 machine minutes: 500 hours Small, 250 Medium, 125 Large
Lite’s 30,000 minutes last 125 hours on the default Large machine.

Past the allowance, machine time costs US$0.10 per 60 machine minutes, paid from credits. At that rate Lite’s minutes are worth US$50, Pro’s US$250 and Max’s US$500. Credits come from the plan’s model usage first, then from any balance you add, with top-ups from US$5. A review’s machine is part of the review and doesn’t use machine minutes.

What a review credit buys

A Low-effort review costs 1 credit, Medium costs 5 and High costs 10. Medium is the default. So Lite’s 250 credits cover 50 reviews and Max’s 2,500 cover 500, not 2,500. The post’s number is the low-effort count, and at High effort Max covers 250. Free accounts get 100 credits a month, or 20 Medium reviews. A covered review includes its model and its machine.

Capy Max's 2,500 review credits: 2,500 low, 500 medium, 250 high effort reviews
Max’s 2,500 credits are 500 reviews at the default Medium effort.

Against Amp

My 14 November comparison found Amp’s US$20 Megawatt plan gave far more cloud machine time. Amp’s pricing page lists 45,000 minutes of Amp-hosted orbs, which is 750 hours of its 2 CPU, 4 GB size, or 1,500 CPU-hours. Capy’s 30,000 minutes at 1 CPU a minute are 500 CPU-hours. On that measure the gap drops from 7.5 times to 3 times. Amp keeps orb time apart from agent usage, and Capy now keeps machine minutes apart from model usage.

The free models

The post also lists three unlimited free models, Particle-1, DeepSeek V4.1 Flash and GLM 5.3 Flash, with no subscription needed. Capy’s pricing page doesn’t mention them or any limit. Capy’s models page lists Particle-1 at US$0 per million tokens. When I checked on the morning of 18 November in New Zealand, it still listed DeepSeek V4.1 Flash at US$0.30 in and US$1.20 out per million tokens, and GLM-5.3-Flash at US$0.15 in and US$0.50 out.

Particle-1 is a coding model from Particle, post-trained on DeepSeek V4.1 Flash. Capy added it on 29 October at US$0.25 in and US$1 out per million tokens.

If you’re already on Capy

Free and Lite organisations move to the new plans automatically. Pro and Max subscribers keep their current plan and monthly credits. The app offers the new plan at the same price, and you can switch whenever you like.

New Lite customers pay US$1 for the first 7 days with a verified card, then US$20 a month. The intro is one per card and per organisation, on monthly Lite only. Capy is also giving a month of Lite to people who comment on its post.


Discover more from Sulat AI & Tech News

Subscribe to get the latest posts sent to your email.