PRICING
Start with 100,000 free tokens. Then buy a pack that never expires, or subscribe and keep every token you don’t use. A $9.90 pack covers about 100,000 short Jev requests.
No subscription and no deadline. Larger packs lower your cost per decision.
$0
100,000 tokens once · ≈500 requests
Unlimited Flows and large batches need any purchase
Try for freeFor kicking the tires
$4.90
8M tokens · one-time purchase
≈ 40,000 requests
$0.613 per 1M input tokens
For building a prototype
$9.90
20M tokens · one-time purchase
≈ 100,000 requests
$0.495 per 1M input tokens
19% lower cost per token than Icebreaker
For a first feature in production
$19.90
50M tokens · one-time purchase
≈ 250,000 requests
$0.398 per 1M input tokens
35% lower cost per token than Icebreaker
For a small app with steady traffic
$49
180M tokens · one-time purchase
≈ 900,000 requests
$0.272 per 1M input tokens
56% lower cost per token than Icebreaker
For batch jobs and several features
$99
450M tokens · one-time purchase
≈ 2,250,000 requests
$0.220 per 1M input tokens
64% lower cost per token than Icebreaker
Buying any pack or plan, even the smallest, lifts the free-account limits for good.
| Free account | After any purchase | |
|---|---|---|
| Saved Decision Flows | 1 | Unlimited |
| Batch rows per upload | 50 | 1,000 |
| Decision Flow billing | Executed branches only | Executed branches only |
| Run saved Flows from the API | ✓ | ✓ |
| Console, batch and API | ✓ | ✓ |
Compare a general-purpose LLM prompt-and-parse step with Jev single calls and with a Decision Flow that only runs the branches each item reaches.
Assumptions: each Jev question ≈ 200 input tokens (output is free); each LLM call ≈ 700 input tokens of instructions, examples and text plus 60 output tokens of JSON; 30 days a month. Your real usage is what the model reports.
A direct model account bills raw tokens for single calls. A JevModel balance pays for the model and everything around it.
The Playground, the visual Decision Flow builder, the Batch CSV/JSONL evaluator and the API all run on one key and one balance.
Packs start at $4.90 and never expire. Subscribers keep unused tokens; yearly plans deliver every token up front.
Failed requests are refunded automatically, and Decision Flows bill only the branches each item actually reaches.
3 runs without an account, 100,000 tokens after you verify an email, and no separate billing setup with the model provider.
To be straight with you: if you only need raw single calls at very high volume, a direct provider account can have a lower per-token rate. Our price covers the hosted tools, batch evaluation, refunds and support on top of model access.
Signup tokens, packs and subscription tokens sit in one balance and never expire. Each successful Playground, batch, Flow or API request deducts the input tokens the model reports; output is free. Requests that fail before a result is saved are refunded. Runs are subject to a site-wide daily capacity limit.
Request counts assume about 200 input tokens per short text question (state plus one question). Longer inputs and more questions use more tokens; the charge is always the model-reported usage.
These are JevModel service prices, separate from the model provider’s direct rate. View model pricing ↗
Need API access? Create a key in account settings and call POST /api/v1/systemone.
Yes. The homepage and Jev playground demo run 3 times per UTC day without an account. Verify an email and you get 100,000 free tokens once, about 500 short requests, plus 1 saved Decision Flow and batches of up to 50 rows.
The API uses the same balance as the Playground and charges only the input tokens the model reports; output is free. For example, the $9.90 pack gives 20M tokens, about 100,000 short requests ($0.49 per 1M tokens).
No. Token packs never expire. Unused subscription tokens roll over month to month, yearly plans deliver all 12 months of tokens up front, and everything you bought stays in your balance if you cancel.
The cards assume about 200 input tokens per short text question. Longer input and more questions use more tokens; the actual model usage determines the charge.
Any purchase lifts the free-account limits: unlimited saved Decision Flows instead of 1, and batch uploads of 1,000 rows instead of 50.
Only the question nodes on the path an item actually takes are run and billed. Branches that are not reached cost nothing, so a deep flow can cost far less than asking every question every time.
No. Each row is a normal Jev run with the same token rules as the workbench and API.
You can. A direct account bills raw tokens for single calls. JevModel adds the visual Flow builder, batch evaluation and one API key on a balance with no monthly commitment, refunds for failed requests and tokens that never expire.
Secure checkout is provided by Stripe. Prices are in USD.
JevModel is an independent service and is not affiliated with TypeSafe AI.
