Sim's free LLM API pricing calculator prices 125 models in dollars, and its own FAQ asks "How do I control LLM costs in an AI agent?". Yet the answer never mentions Sim. There are no credits, no plan, no spend limit and no BYOK, only the site-wide "Start building" button. Someone who has just modelled 1,000 requests a day is a qualified buyer with a budget question, and the page leaves the Sim half of the sum to them. Sim's answer is unusually concrete: 1 credit ($0.005) per run as the base charge, model usage from recorded tokens, BYOK to pay the provider directly, and spend caps per member.
First action
Write the formula for runs, base credits, model credits and plan fit, and test it on three example workloads.
How to measure it
Calculator-to-sign-up rate
- Now
- –
- Target
- 2x
- By
- 60 days
Steps
Agree the math
Runs × 1 credit, model tokens at the Sim rate, BYOK mode, overage at $0.005, as in the costs doc.
Founding Marketing 1 day
Map plans
Thresholds for Free, Pro, Max and an Enterprise prompt above Max.
Founding Sales 1 day
Design the panel
One line in credits, one plan badge, a BYOK toggle and a batching tip under the dollar result.
Founding Design Engineer 2 days
Carry the budget
Pass model and monthly budget as sign-up parameters and pre-fill the usage cap.
Founding Design Engineer 3 days
Reuse it
A compact panel on the token counter and the model pages.
Founding Design Engineer 3 days
Instrument
Events for panel views, BYOK toggles, button clicks, sign-ups and first run.
Founding Marketing 1 day
Route big budgets
Estimates above the Max allowance offer a call with sales.
Founding Sales 1 day
Review
Compare sign-up and first-run rates with the month before.
Founding Marketing monthly
Why it works
Under the dollar result, add a "Run this on Sim" panel. (1) Monthly runs × 1 credit base charge, plus hosted model usage converted at $0.005 a credit, with a BYOK toggle that shows the base charge only. (2) The plan that covers it: Free (1,000 one-time credits), Pro ($25 per user, 6,000 credits a month plus 2,000 a week) or Max ($100 per user, 25,000 plus 4,000 a week), and any overage. (3) A note that several model calls inside one run share one base charge, so batching lowers cost. (4) A "Start with this budget" button that carries the model and monthly figure into sign-up and pre-fills the workspace usage cap. Be honest at volume: at the default 1,000 requests a day, if each request is its own run, the base charge alone is about 30,000 credits ($150) a month, which is more than Pro's allowance. Showing that up front, with BYOK and batching as the levers, builds more trust than a surprise bill. Add a compact version of the panel to the token counter and the 400+ model pages.
Cost is the first objection to agents, and this page attracts people actively working on it. Answering in Sim's own units turns a research tool into a sales tool, and it shows spend limits (a headline feature) at the moment they matter. It gives the founding marketer a clean activation metric: calculator, then sign-up with a budget set, then first run.
Ready to use
Copy"Run this on Sim" panel
Inputs (already on the page): input tokens, output tokens, % cached, requests/day, model runs_month = requests_day × 30 (editable: calls per run) base_credits = runs_month × 1 model_credits = model_cost_usd ÷ 0.005 (0 when BYOK is on) total_credits = base_credits + model_credits Plan: ≤ 1,000 one-time → Free; within Pro allowance → Pro $25/user; within Max allowance → Max $100/user; above → Enterprise call Overage = credits above allowance × $0.005 Button: "Start with this budget" → /signup?model=<id>&budget=<usd> Tip: "Several model calls inside one run share one base charge."
Risks and how to handle them
The honest base-charge figure scares away high-volume users
Show the BYOK and batching levers next to it and offer a sales call above Max.
The estimate drifts from real bills
Use the same rate table as billing and date it, as the calculator already does.
Evidence
- Your websiteLLM API pricing calculator ("Free tool", 125 models, prices updated November 5, 2026): estimates cost per request, day and month in US dollars from tokens and requests per day (default 2,000 input and 500 output tokens at 1,000 requests a day). It never mentions Sim credits, plans, spend limits or BYOK; its FAQ asks "How do I control LLM costs in an AI agent?"; the only Sim calls to action are the site-wide "Start building" and "Request a demo".Source
- Your docsCredits, billing and limits: each billable run costs 1 credit ($0.005), and child executions share the parent's base charge; hosted model calls are billed from recorded token usage; with your own provider keys (BYOK) you pay the provider directly and the base run charge still applies; paid usage is capped at the included allowance by default, and "Manage credits" sets a member's cap; a Pro account with 1,000 credits of overage is billed $5 on top of $25.Source
- Your websitePricing page: Free $0 with 1,000 one-time credits, 1 workspace, 2 concurrent executions and 30-day logs; Pro $25 per user a month with 6,000 credits a month plus 2,000 a week; Max $100 per user a month with 25,000 credits plus 4,000 a week; Enterprise custom. "On Sim Cloud, one credit is worth $0.005." An "Annual -15%" toggle, and no FAQ section.Source
- HiringFounding Marketing (San Francisco, posted September 24, 2026): "You'll own how people discover, understand, and adopt Sim" through positioning, technical content and demos, launches and channel tests, measured on activation and qualified pipeline rather than attention.Source