Every AI Excel add-in gets compared on features. The difference that actually decides your month is structural: Claude, ChatGPT, Shortcut, and Copilot resell frontier-lab inference, so they price a seat around it. EBITDAI runs on DeepSeek and Kimi, whose tokens cost a fraction of that, so a smaller bill buys as much modeling or more: $5 or $15/month, and no cap of our own on top of it.
Plan figures are as advertised by each provider in August 2026 and can change on their side.
Same question for every column: what does a serious month of modeling cost, and what stops you mid-build?
| EBITDAI + Kimi | Claude for Excel | ChatGPT for Excel | Shortcut | Copilot | |
|---|---|---|---|---|---|
| Monthly cost to use seriously | $34 all-in, flat: the $15 Pro plan plus a $19 Kimi Code subscription, and Pro is the plan that adds the QuickBooks, Campfire and Stripe connectors | $20 Claude Pro minimum; serious modeling pushes to Max at $100 or $200/month | $20 Plus | $125/month billed monthly; the advertised $100/month requires prepaying a full year | $30 per seat per month, on top of Microsoft 365 |
| What that buys (usage) | Model all day, no meter: roughly 300 to 1,200 calls per 5-hour window even on Kimi's tier below Moderato (per plan trackers), zero per-token charges, no cap from EBITDAI | A pool shared with all your other Claude use, reported at 45 messages per 5-hour window; reviewers describe burning Pro limits in minutes on heavy modeling | A shared agentic credit pool; a typical Excel task costs 5 to 20 credits | 1,000 monthly credits, at 2 to 15 credits per message | Enterprise licensing under Microsoft's terms |
| What happens when you hit the cap | There is no cap from EBITDAI. If you outgrow Kimi's base plan, its tiers scale to 5x, 15x, and 30x base credits | Wait for the reset, or upgrade: Max is $100/month (5x Pro) or $200/month (20x Pro) | Buy extra usage, or wait for the pool to refresh | Pay overage at $0.10 per credit, a rate that doubled in April 2026 | Ask IT: usage is governed by your Microsoft agreement |
| Model choice | Your pick of Kimi k3, DeepSeek V4.1 Flash, DeepSeek V4 Pro and Gemini 3.8 Flash on Pro, plus Meta Muse Spark 1.3 in both tiers, or your own Kimi, DeepSeek or Gemini key on Pro, switchable mid-session | Claude models only | GPT models only | Its own lineup; the free tier is pinned to cheaper models | Microsoft picks the models |
| Free entry | Free plan: 150 AI requests a week on Meta Muse Spark 1.3 (Meta trains on everything you send with it) | None for the add-in: Claude Pro at $20/month is the floor | Yes, a limited free tier | 20 credits per week, pinned to cheaper models | None: enterprise licensing |
All plan prices, allowances, and overage rates as advertised by the providers, August 2026. Claude Pro limits reporting: aitoolsreview.co.uk. ChatGPT credit costs per OpenAI's published guidance. Shortcut credit rates and overage per Shortcut's published pricing. EBITDAI imposes no usage caps of its own; your Kimi plan governs usage.
In our production tests, typical single tasks complete in 24 to 47 seconds, and that holds across both providers we support. A complete DCF builds in under five minutes. For contrast, ChatGPT for Excel users on its Hacker News launch thread report basic styled builds taking 15 to 20 minutes.
The two subscription ladders start at nearly the same price. They end in very different places. Tier math as the providers advertise their plans, August 2026.
A $19 entry point against Claude Pro's $20. But one ladder tops out at 30x base usage for less than the other charges for 20x, and none of it competes with a chat product for the same pool.
This comparison cuts both ways. Each of these tools has a real case:
Because we run on DeepSeek, Kimi, Gemini and Meta Muse, whose token prices are a fraction of the big labs'. Lite is $5/month and carries 100x more usage than free on Meta Muse Spark 1.3 Contributor; Pro is $15/month and carries 7.5x more usage than lite across every included model, so heavy Excel work sits well under a $20 Claude Pro floor. We publish the included usage as a multiple of the plan below it, measured the way the pricing page explains. Prefer to run on your own provider account? Pro ($15/month) runs on your own keys: your AI is billed by your provider at its own advertised price, with zero markup from us. On your own key, DeepSeek and pay-as-you-go Kimi calls go straight from the add-in to the provider; Kimi Code and Gemini cannot be called from a browser, so those pass through our stateless relay, which stores nothing.
Your Kimi Code plan governs them. The base subscription is $19/month flat with no per-token charges, and Kimi's paid tiers scale to 5x, 15x, and 30x base credits if you need more headroom. In our experience the base tier covers heavy daily modeling. EBITDAI is currently the only Excel add-in that runs on a Kimi Code subscription at all.
EBITDAI adds no cap of its own beyond the free plan: Free includes 150 requests a week, and Lite and Pro have no weekly limit, so your provider's plan is the only thing that governs usage there. On Kimi that means a flat subscription allowance; on DeepSeek it means pay-per-token with no ceiling, where a full model build costs a few cents. What you never get from us is a second meter on top of the one you're already paying for.
The $5/month Lite plan carries 100x more usage than free on Meta Muse Spark 1.3 Contributor with no weekly request cap, so light regular use is $5 a month all-in. If you outgrow it, Pro takes your own DeepSeek key, so you can pay DeepSeek pay-per-token directly: that lands near $15 plus a few dollars of usage.
From $5/month, a quarter the price of a $20 Claude plan. Or $34/month all-in with a Kimi Code subscription, flat, for all-day modeling.
Free plan: 150 AI requests a week on Meta Muse Spark 1.3, and Meta trains on everything you send with it. No credit card.