Start free. Upgrade when you want more.
Try EBITDAI
No card required.
Lowest Price Excel Agent on the Market
Per month. $48 billed annually.
Everything in Free, plus:
Everything EBITDAI can do
Per month. $144 billed annually.
Everything in Lite, plus:
Secure checkout by Stripe · cancel anytime from your account page.
Free | Lite | Pro | |
|---|---|---|---|
| Models and usage | |||
| Included models | Meta Muse Spark 1.3 only (Meta trains on everything you send with it) | Meta Muse Spark 1.3 only, the same Contributor tier as the free plan (Meta trains on everything you send with it) | Kimi k3, DeepSeek V4.1 Flash, DeepSeek V4 Pro and Gemini 3.8 Flash, plus Meta Muse Spark 1.3 in both tiers: not used for training, or trains on your data |
| Included usage every month | 150 AI requests a week on Meta Muse Spark 1.3 | 100x more usage than free | 7.5x more usage than lite |
| Weekly request cap | 150 requests a week | None | None |
| With your own Kimi Code subscription$19/month flat, bought from Moonshot. No per-token charges. | Not included | Not included | Flat $19/month, no per-token charges |
| Usage dedicated to Excel, never shared with chat or coding | Included | Included | Included |
| No 5-hour windows, no weekly lockouts | Not included | Included | Included |
| Switch models in the pane, mid-session | Not included | Not included | Included |
| Extra-usage packs$5, $10 and $25 packs. Never expire. On Lite they are spent on Meta Muse Contributor only. | Not included | Included | Included |
| Connectors | |||
| QuickBooks, Campfire and Stripe connectorsPull live P&L, balance sheet, cash flow, invoices, customers and more straight into the workbook. | Not included | Not included | Included |
| Privacy and your own keys | |||
| Own API keysZero markup. You pay your provider its published prices. Pro connects your own key for every provider: Kimi Code (subscription), Moonshot (metered Kimi), DeepSeek and Gemini. Lite and the free plan run on included usage only and cannot connect a key. | Not included | Not included | Kimi Code, Moonshot, DeepSeek and Gemini |
| Workbook data path on your own keyNot every provider can be called from a browser. DeepSeek and pay-as-you-go Kimi (Moonshot) return CORS headers, so the add-in talks to them directly from your device and nothing transits an EBITDAI server. Kimi Code and Gemini do not, so those calls pass through a stateless EBITDAI relay that forwards your key per request and stores no prompt or workbook content. | Not included | Not included | Browser-direct on DeepSeek and metered Kimi; stateless relay for Kimi Code and Gemini |
| Usage limited only by your own provider | Not included | Not included | On your own key |
| EBITDAI never stores your prompts or workbook content | Not included | Included | Included |
| Conversations logged and read by the EBITDAI team to improve the productFree plan only: your prompts, the analyst's replies and the workbook context you share are kept for 90 days and reviewed by us. Lite is a paid plan, so its turns are not logged. | Included | Not included | Not included |
| The analyst itself | |||
| The EBITDAI analyst in Excel for Windows, Mac and web | Included | Included | Included |
| Full 3-statement modeling doctrine | Included | Included | Included |
| Debt, working capital, capex, SaaS and waterfall playbooks | Included | Included | Included |
| Audit and fix an existing workbook, not just build a new one | Included | Included | Included |
| File attachments and screenshots in chat | Not included | Included | Included |
How the usage comparison is measured. A complete model build measures 716K input tokens at 97% cache hits plus 10K of output in our published benchmark. On Meta Muse Spark 1.3's Contributor rate card, the cheapest model in the plans, that is $0.0055 a build, and at published DeepSeek V4.1 Flash off-peak prices it is $0.0113, so what an allowance buys moves with the model you pick. For comparison, a $20 Claude Pro plan's entire monthly allowance goes to Claude for Excel running Claude Fable at the top of its estimated range, which is a smaller monthly capacity for a higher price. The Kimi Code option is a $19/month flat-rate subscription you buy from Moonshot and connect on Pro, on the same benchmark's measured calls per build. Neither Anthropic nor Moonshot publishes exact quotas, so the Claude and Kimi Code numbers are estimates. One difference no estimate covers: on Claude that allowance is shared with chat and coding, while yours is dedicated to Excel.
The free plan needs no card. Install the add-in, point it at a workbook you already have, and ask it to find why your statements do not tie. If it earns a place in your workflow, pick a plan then. See how the add-in works.
Lite comes with 100x more usage than free, on Meta Muse Spark 1.3 on Meta's Contributor tier, and no weekly request cap: spend it in one sitting if you like. Pro comes with 7.5x more usage than lite, across every included model, one shared wallet you can reload anytime: Kimi k3, DeepSeek V4.1 Flash, DeepSeek V4 Pro and Gemini 3.8 Flash, plus Meta Muse Spark 1.3 in two tiers, the standard one that Meta does not use for training and the Contributor one where Meta trains on everything you send with it. Either way your usage is dedicated to Excel rather than shared with chat and coding.
The free plan (150 requests a week on Meta Muse Spark 1.3, no card) is for trying EBITDAI. Lite ($5/month) is the lowest price paid plan: the same Meta Muse Spark 1.3 Contributor model as the free plan, with 100x more usage than free, no weekly request cap, and file attachments and screenshots. Pick Pro ($15/month) for every included model: Kimi k3 (the model that matched Claude for Excel in our published benchmark), DeepSeek V4.1 Flash, DeepSeek V4 Pro and Gemini 3.8 Flash, plus both Meta Muse Spark 1.3 tiers, on one shared allowance carrying 7.5x more usage than lite, with your own API keys for every provider (Kimi Code, Moonshot, DeepSeek and Gemini) and the QuickBooks, Campfire and Stripe connectors.
Models, usage and connectors. Lite runs Meta Muse Spark 1.3 on Meta's Contributor tier only, with 100x more usage than free, no weekly request cap and no own keys. Pro ($15/month) adds the other four included models, a monthly allowance across every model carrying 7.5x more usage than lite, your own API keys for Kimi Code, Moonshot, DeepSeek and Gemini, and the QuickBooks, Campfire and Stripe connectors. Both include the full analyst: the modeling doctrine, the audit path, and file attachments and screenshots.
They are now Lite and Pro. The old $5 plan is Lite, at the same $5/month, now running Meta Muse Spark 1.3 on Meta's Contributor tier with 100x more usage than free, no weekly request cap, and attachments. The old $15 and $25 plans merged into Pro: every model they had, a monthly allowance carrying 7.5x more usage than lite, your own API keys for every provider, and the QuickBooks, Campfire and Stripe connectors that used to be $25 only, all at $15/month. Existing subscribers keep their plan history and were moved to the new plans; nothing to do. The changelog has the full migration note.
No lockouts. On Claude, running out means waiting for the clock to reset or jumping from $20 to $100. On EBITDAI, your usage is dedicated to Excel, you can see how much you have left in the pane, and if you run out you buy more on the spot in about a minute. Free includes 150 requests a week, and Lite and Pro have no weekly limit. Running on your own key goes further: on Pro, with your own Kimi, DeepSeek or Gemini key, usage is whatever your provider allows.
You will see your remaining usage in the pane. Run out before the month resets and you can buy extra usage on the spot: $5, $10, or $25 packs, right in the pane or from your account. Extra usage never expires, and your monthly usage refreshes either way. On Lite, extra usage is spent on Meta Muse Contributor.
Straight answer: where your data goes follows the model you pick, not the plan alone. On Lite, and on the free plan, the only model is Meta Muse Spark 1.3: Meta serves it on the Contributor tier, so your prompts, the replies and every workbook value the analyst reads go to Meta Platforms, Inc. in the United States, and Meta trains on them under its published Contributor terms. That is true on Lite even though it is a paid plan. On Pro, included usage passes through an EBITDAI relay to the provider that serves the model you picked: DeepSeek V4.1 Flash and DeepSeek V4 Pro to Hangzhou DeepSeek Artificial Intelligence Co., Ltd., incorporated in the People's Republic of China and processing there; Kimi k3 to Moonshot AI PTE. LTD., incorporated in Singapore; Gemini 3.8 Flash to Google LLC and Meta Muse Spark 1.3 to Meta Platforms, Inc., both incorporated in the United States. Pro includes all five. The relay stores no prompt or workbook content, and EBITDAI never stores your workbook on any paid plan. One exception: free-plan conversations (your prompts, the analyst's replies and the workbook context you share) are logged and read by the EBITDAI team to see what works and what frustrates, and deleted after 90 days; Lite turns are not logged, because it is a paid plan. Meta Muse Spark 1.3 also comes in two tiers, the Contributor tier (the only tier on the free plan and on Lite, and the option shown as "trains on your data"), where Meta trains on everything you send with it, and the standard tier on Pro, which Meta does not use for training under its published pricing terms. If your data must not go through us at all, run on a key of your own on Pro: Pro takes your own Kimi Code, Moonshot, DeepSeek or Gemini key. Own-key DeepSeek and own-key Moonshot (pay-as-you-go Kimi) calls go straight from your browser to the provider and never touch our servers; own-key Kimi Code and own-key Gemini cannot be called from a browser, so those pass through a stateless relay that stores no prompt or workbook content. The full data path is documented at ebitdai.co/security.
Lite and the free plan run Meta Muse Spark 1.3 on Meta's Contributor tier only. Pro runs Kimi k3, DeepSeek V4.1 Flash, DeepSeek V4 Pro and Gemini 3.8 Flash, switchable mid-session, plus Meta Muse Spark 1.3 in two tiers, the standard one that Meta does not use for training and the Contributor one where Meta trains on everything you send with it. Pro also runs your own Kimi key (flat-rate Kimi Code or metered Moonshot), DeepSeek key or Gemini key. There is no premium tier for better models: Pro is the plan with every model.
Yes. EBITDAI is currently the only Excel add-in that runs on a flat-rate Kimi Code plan. On Pro, that setup covers unlimited modeling with no per-token charges. On the Kimi Code waitlist? A pay-as-you-go key from the Moonshot Open Platform runs the same k3 model with no waitlist, and Pro takes it too.
Pay yearly and every paid plan is 20% off: Lite is $48/year (works out to $4/month) and Pro is $144/year ($12/month). The free plan stays free.
Yes. Manage or cancel your subscription from your account page in one click.