Pricing

Smarty Pants pricing

Pay only for what you use, from a balance you top up. No subscription. Using your own Macs is free.

Short answer: Starting is free: work on Macs you or your Group own costs nothing, and new accounts get $10 of credit for 90 days. Paid use on other people's Macs is priced at about 95% of the cheapest seller of the same thing elsewhere. Chat models cost from $0.0171 per million input tokens, batches are half price, a 2 vCPU workspace is $0.0129 an hour billed per second,. Checked November 19, 2026.

What does each product cost?

One balance pays for everything. Each product has its own page with limits and code.

ProductPriceWho can use it today
AI models (chat)Per million tokens, from $0.0171 in / $0.0855 outAnyone with an API key
BatchesHalf the normal per-token price, results within 24 hoursAnyone with an API key
Embeddings and reranking$0.01 and $0.00955 per million input tokensAnyone with an API key
Speech to textAbout $0.011 an hour of audio, billed per secondIn testing
Pictures$0.003 a pictureIn testing
Video$0.011 a 480p clip (2, 3 or 5 seconds)Anyone with an API key
Code runs and workspaces$0.0129 an hour for 2 vCPU and 2 GB, billed per secondFree on your own Macs; paid use on our own account first
Groups (your team's own Macs)Free for members on the Group's MacsInvite only
Mac app buildsFree, on your own or your Group's Macs onlyComing (switched off today)

Starter credit: $10 for a newly verified person or agent, once, usable for 90 days. Top-ups start at $10 and are paid on Stripe. Live prices: /v1/models and /v1/prices/workspaces.

What is free, and what do you pay for?

We give away the places people start and charge close to market for the work that grows with use.

FreePaid, at about 95% of the cheapest seller
Workspaces, code runs and model calls on Macs you ownThe same work on Macs other people own
Everything a Group member runs on the Group's MacsBig-model output and long agent runs
$10 of starter credit for 90 daysPictures, video, and models we offer first
Requests we can't serve, workspaces that never start or that we loseBatches (half the normal per-token price)

Group seats (coming)

Seats for teams are planned and not on sale yet. What we expect to offer:

  • Entry, free per person: the Group's shared folder, task board and history, unlimited agents, free use of the Group's own Macs, and Qwen3.5 9B free for up to 200,000 tokens a day per person.
  • Pro, about $49 per person a month with $50 of credit: always-on workspaces, more saved storage, steady Macs first, and a discount on use past the credit.

Seat names, prices and limits may change before they open. Use beyond your Group's Macs is paid at the network price, with a spending cap you set.

How do model prices compare with the cheapest seller?

For a model other providers also sell, we take the cheapest seller we can find and charge about 95% of its price. The builds are not identical (ours are 4-bit MLX builds on Apple silicon, named in build_id), so compare quality on your own examples too.

ModelSmarty Pants, per million tokensCheapest seller elsewhereOurs vs. theirsStatus
Qwen3.5 9B $0.076 in / $0.114 out $0.08 / $0.12 (DeepInfra) 95% Released
Qwen3.8 27B $0.037 / $1.19 $0.039 / $1.25 (Chutes) 95% Released
Qwen3.6 35B-A3B $0.048 / $0.665 $0.05 / $0.70 (Darkbloom) 95% Released
gpt-oss-20b $0.0171 / $0.0855 $0.018 / $0.09 (Darkbloom, AkashML) 95% Released
Muse Glimmer 30B $0.285 / $1.14 $0.30 / $1.20 (DeepInfra, ModelRun) 95% Released
Qwen3 Reranker 0.6B $0.0095 in $0.01 in (DeepInfra (9 Nov)) 95% Released
Whisper large-v3 turbo about $0.011 an hour of audio $0.012 an hour ($0.0002 a minute) (DeepInfra) 95% In testing, opens soon
Wan 2.2 5B video, 480p clip $0.011 a clip about $0.0113 a clip (half of $0.0225 for 720p) (DeepInfra) 98% Released
FLUX.1 schnell picture $0.003 a picture about $0.003 to $0.005 a picture (fal (reported)) about 100% In testing, opens soon

Models only we sell

When nobody else sells a model yet, we set the price from the nearest comparable model.

ModelSmarty Pants, per million tokensNote
Arete 27B$0.20 / $1.20No other seller found on 19 Nov 2026; priced like a 27B chat model.
Axion 1.0 35B$0.09 / $0.80No other seller found; priced like its 35B-A3B base at launch.
Nanbeige4.3 33B$0.30 / $2.00No other seller found.
Qwen3 Embedding 0.6B$0.01 inNo current seller price verified.
Qwen3.5 4B$0.10 / $0.25In testing.

Seller prices from each linked page on the date shown, mostly OpenRouter's list of sellers on November 19, 2026. Picture prices come from a secondary source and may be out of date. Model price calculator.

How do workspace prices compare with other sandboxes?

A Smarty Pants workspace is a sealed Linux VM with 2 vCPU and 2 GB of memory, billed per second from when it is ready until it ends. Code runs (POST /v1/code/run), fleets and coding-agent jobs use the same rate per workspace. Here is the same size machine elsewhere, per hour.

ServicePer vCPU-hourPer GB-hour2 vCPU + 2 GB, per hourFree credit
Smarty Pants workspace$0.0043$0.00215$0.0129$10 starter credit for 90 days; free on your own and your Group's Macs
boat$0.0045$0.00225$0.0135Not checked
Northflank$0.01667$0.00833$0.05None stated
Fly.io Sprites$0.0385$0.021875$0.12None stated
E2B$0.0504$0.0162$0.1332$100 one-time credit (Hobby)
Daytona$0.0504$0.0162$0.1332$200 starter credit
Modal Sandboxesabout $0.071 ($0.1419 a core of 2 vCPU)$0.024$0.1899$30 a month (Starter)
Vercel Sandbox$0.128 (active CPU)$0.0212$0.2984Pro plan credit

Other services' prices are from the linked comparison pages on their stated dates; check each provider before you decide. Prices are for running time only; storage, network and plan fees differ by provider.

What the lower price does not buy you

  • No GPUs and no x86 machines: workspaces are ARM64 Linux on Apple silicon.
  • Capacity depends on Macs that are idle and online. A Mac with a workspace booted and waiting starts one in about a second; otherwise it takes longer, and sometimes no Mac is free.
  • The Mac's owner comes first. If they come back, the workspace ends within 5 seconds (save it, or turn on failover, to carry on elsewhere).
  • Workspace contents are not confidential from the Mac's owner. Use your own or your Group's Macs for private work.
  • No SOC 2 report yet. Paid workspaces on the general network are on our own account first.

What is free?

Anything that runs on Macs you own is free, and so is anything a Group member runs on the Group's Macs. Add where: "mine" or where: {group: "<Group id>"} to a workspace request. You are also never charged for a request Smarty Pants can't serve, a workspace that never became ready, a workspace we lost, or time after you end one.

Compare us with others on the comparison pages, or read LLM API pricing explained.

Frequently asked questions

How does Smarty Pants set its prices?

The places you start are free: your own Macs, your Group's Macs and the $10 starter credit. Paid use on Macs other people own is priced at 95 to 100% of the cheapest price we can find for the same model or the same kind of machine elsewhere. We checked on November 19, 2026; most prices are 95% of the cheapest seller that day.

Is there a subscription or a minimum?

No subscription. You add credit to a balance (top-ups start at $10) and each request is paid from it. New accounts get $10 of free credit, usable for 90 days.

What am I never charged for?

A request Smarty Pants can't serve, a workspace that never became ready, a workspace our side lost (its Mac went quiet), and time after you end a workspace.

Is using my own Macs really free?

Yes. Workspaces, code runs and model calls on Macs you own, or on your Group's Macs for its members, cost nothing. You pay only when work runs on Macs other people own.

Where is the live price?

GET https://smartypants.network/v1/models lists every model's current price, and GET https://smartypants.network/v1/prices/workspaces lists the workspace price. Prices on this page were checked on November 19, 2026 and can change.

Are you cheaper than everyone?

Not always, and not forever. We aim for a little under the cheapest seller on the day we check, but other providers change prices often, and some offer things we don't (GPUs, longer context, guaranteed capacity, SOC 2 reports).

Try a pinned open model on real Macs.

See every model, its listed price and the code to call it. Smarty Pants works with the OpenAI SDK you already use: change the base URL and the key.

Requests run on Macs owned by other people. Owners agree not to read requests and the app doesn't show them, but a serving Mac decrypts each request in memory to answer it, so that is a rule and a design, not a technical guarantee. Don't send sensitive data on the general network.