EBITDAI
    The only Excel add-in that runs DeepSeek

    DeepSeek inside Excel.

    A complete financial model costs a few cents on DeepSeek's rates. That is why the Pro plan's included monthly AI usage across every model, 7.5x what Lite carries, goes a long way here, without the meter Claude and ChatGPT add-ins put on you.

    What your modeling actually costs

    Denominated in DeepSeek V4 API rates and EBITDAI's measured cost per complete model build ($0.06).

    On the EBITDAI Pro plan
    $15/mo all-in
    7.5x more usage than Lite across every model, dedicated to Excel
    Your own DeepSeek key
    ~$1.20
    pay-as-you-go at DeepSeek's rates
    Same work at Claude API rates
    ~$24.00
    Sonnet API pricing, same token volume
    Per 1M tokensDeepSeek V4-FlashClaude Sonnet API
    Input$0.22 - $0.44$3.00
    Output$0.66 - $1.32$15.00

    Published API rates as of August 2026; DeepSeek shows off-peak/peak. Roughly 14-23x less per token.

    Two ways to run it

    Included (default)

    Pro is $15/month, all-in: sign in and start modeling, remaining usage visible in the pane. 7.5x more usage than Lite across every model, and a complete model build on DeepSeek V4.1 Flash costs a few cents of it.

    Your own DeepSeek key (built into Pro)

    Connecting your own DeepSeek key is included on the Pro plan. Requests on that key go browser-direct to DeepSeek and DeepSeek bills you at its rates, while the plan's included usage stays available whenever you switch back. Key setup guide.

    Questions

    Can I use DeepSeek in Excel?

    Yes, through EBITDAI. DeepSeek has no first-party Excel add-in; EBITDAI is the only agentic Excel add-in that runs DeepSeek's models, either on an EBITDAI plan or on your own DeepSeek API key, which the Pro plan supports.

    What does a financial model cost to build on DeepSeek?

    Measured on real builds: a complete from-scratch model (for example a 6-year DCF) consumes a few cents of DeepSeek usage. The $15/month Pro plan includes 7.5x more usage than Lite across every model, and a complete build on DeepSeek V4.1 Flash costs a few cents of it.

    Why is DeepSeek so much cheaper than Claude or OpenAI?

    DeepSeek publishes API rates of $0.22-0.44 per million input tokens and $0.66-1.32 per million output tokens (V4-Flash; the spread is off-peak versus peak hours), versus $3 and $15 for Claude Sonnet's API: 14-23x less per token. EBITDAI passes that economics through instead of reselling expensive lab tokens.

    Is DeepSeek good enough for financial modeling?

    EBITDAI ships with a mechanical eval harness: models must produce error-free, correctly structured workbooks (three-statement builds, debt schedules, working capital, DCFs) before we ship them in a plan. DeepSeek V4 passes the full suite, including building a correct Tesla DCF from a blank workbook.

    Where does my workbook data go?

    On a plan's included usage, requests route through EBITDAI's relay to DeepSeek, which processes them in the PRC; the relay stores nothing on paid plans, while free-plan conversations are logged and reviewed to improve the product. With your own DeepSeek key, which the Pro plan supports, your workbook goes directly from your machine to DeepSeek and never touches EBITDAI's servers. Full details on the Security page.

    Frontier modeling at DeepSeek prices.

    Pro is $15/month. Begin for free today; cancel anytime.

    Begin for free today