Speed

10× faster than a frontier LLM

Median, vs Claude Opus 5 on the same decisions.
Show the mathHide the math

Full method and raw figures↗

Cost

Over 100× cheaper than a frontier LLM

Per decision, on the same questions.
Show the mathHide the math

Full method and raw figures↗

Intelligence

More accurate than other decision models

It reasons through hard questions, which no other decision model does, and reads images.
Show the mathHide the math

Every run, published↗

• 75× cheaper document extraction than Opus 5.5
• 2.0–2.3× cheaper than Haiku 5.5
• 3× faster streaming decisions
• 150 ms extraction

Free

$0
Includes $1 of usage a month
≈ 50K short questions
≈ 2.2K invoices extracted
Input $0.05/ 1M tokens

Developer

$14/mo
Includes $14 of usage a month
≈ 700K short questions
≈ 31K invoices extracted
Input $0.05/ 1M tokens

Starter

$49/mo
Includes $49 of usage a month
≈ 2.7M short questions
≈ 110K invoices extracted
Input $0.046/ 1M tokens

Pro

$99/mo
Includes $99 of usage a month
≈ 5.9M short questions
≈ 220K invoices extracted
Input $0.042/ 1M tokens

• Priority support

Growth

$249/mo
Includes $249 of usage a month
≈ 16M short questions
≈ 550K invoices extracted
Input $0.038/ 1M tokens

• 128K context

• Priority support

• All plans

• Output $1 / 1M tokens

• Web search $0.10 each

• No seats

• No overage invoices

Decision

Decides.

Text and a question in. A typed answer out.
This is the third time my order is late.
Is the customer upset? Yes
$0.00002 a question

Streaming

The answer updates as the text arrives.
Hmm. 72 months feels long.OK, 48 months then. Send me the quote.
Agreed to go ahead?No Yes
$0.0013 a live call, four updates

Extraction

Words in. Data out.
Two oat lattes and a croissant, pickup at 8:15 for Dana.
item Two oat lattes_item_ a croissant_time_ 08:15_customer_ Dana
$0.00045 an invoice, four fields

And more

Same key.

Everything below comes with every plan.

• Yes / No
• Choice
• Scale
• Sort
• Tags
• Batching
• Reasoning
• Images
• Web search

Costs at the Developer plan’s rates.

• Short question One decision on a message $0.00002/call
• Long document One decision on 8,000 tokens $0.00042/call
• Five tags Moderate a post $0.0001/call
• Invoice Extract four fields $0.00045/call
• Live call Two questions, four updates $0.0013/call

2 Set your volume
250K calls a month 1K 10K 100K 1M 10M

400 tokens in · 0 out, per call. Costs on the Developer plan’s rates.

Usage a month

≈ $5.00

Fits Developer· $14/mo

• Free over
• Developer 36%
• Starter 9%
• Pro 4%
• Growth 2%

In

$0.038–0.05 per 1M tokens, by plan

Everything Sage reads: your content, questions and options. Images count as tokens.

Out

$1 per 1M tokens

What extraction and streaming send back, plus reasoning. A decision’s answer is free.

Search

$0.10 per search

Only when you turn on web search and it runs.

Every response reports its tokens in meta.usage. POST /usage/estimate returns them before you send a decision.

What happens when my usage runs out?

Calls return 402 until the next period or an upgrade. No overage invoices, ever.

Does unused usage roll over?

No. Each period starts with its plan’s full amount.

Do I pay per seat?

No. Plans have no seats.

Can I see the cost before I send?

Post a decision to /usage/estimate and get its tokens back. After a call, meta.usage has the exact count.

I need more than the largest plan.

Email team@levanto.ai for volume pricing.

Intelligence, at runtime

Costs at the Developer Plan’s rates.

Everything Sage reads: your content, questions, and options. Images count as tokens.