Blueflame AI Pricing and Blueflame Agent Credit Rates

Blueflame AI has no public price list. Since Datasite bought it in July 2025, Blueflame runs inside Grata Intelligence and is sold through Grata's Growth, Scale and Alpha plans, none of which show a dollar figure. What is public is the usage side: Grata's rate card charges the Blueflame Agent 7.50 credits per million input tokens and 50 credits per million output tokens, and the heavier Blueflame Composer five times that. The dollar value of a credit is set in each client's order form. We checked blueflame.ai, grata.com and the rate card in October 2026.

How much does Blueflame AI cost?

Nobody outside a Datasite contract can quote you a sticker price, and anyone who does is guessing. Blueflame's own site has no pricing page (the /pricing address returns a 404), and every button reads "See it in action", which books a demo. Grata's pricing page lists three tiers by name and audience, then ends each one with "Book a demo".

The bill has two layers. The first is the platform subscription, negotiated with Datasite and tied to the Grata tier. The second is usage, measured in credits, and that part is published. Our Grata alternatives and pricing page records what buyers have reported for the subscription layer: a commonly cited range of $10,000 to $100,000 a year, with a widely repeated $155,000 "median" that rests on six buyer reports and should not be trusted.

Two lines in that table matter for budgeting. Scale charges a setup fee for ingestion that Alpha includes, and both meter ingestion in credits on top. So the cheaper tier is not automatically the cheaper contract if you plan to load a lot of your own documents.

What does the Blueflame Agent cost in credits?

Grata publishes a rate card, and it is the only primary pricing document for Blueflame we could find. It says plainly that "Blueflame Agent usage is charged at the applicable rates set forth in this Rate Card", and that when the agent calls Blueflame Composer for advanced work, Composer rates apply to that part of the job.

The rate card also prices third-party models you can switch the agent to, from very cheap small models at under 3 credits per million output tokens up to frontier models at 100 to 250. If you pick a non-default model, its rate replaces the Agent rate, and Composer charges still apply for the capabilities you use. What the page never says is what one credit costs in dollars. It notes that "additional charges may apply" under the client's order form, which is where the credit price lives.

How to estimate a Blueflame bill before the demo

You can size usage in credits now and convert it once sales gives you the credit price. Three worked examples using the published rates:

• A screening question. The agent reads 2 million tokens of company data and writes 100,000 tokens of answer. That is 2 x 7.50 plus 0.1 x 50, or 20 credits.
• The same job through Composer. Identical token counts at Composer rates cost 2 x 37.50 plus 0.1 x 250, or 100 credits. Same question, five times the bill.
• A data room you keep live for a year. Ingesting and storing 10 GB costs 10 x 225, or 2,250 credits, before anyone asks it a single question.

The practical lesson: ask how often the agent escalates to Composer on your typical workflows, and ask for a monthly credit report during the pilot. A team that runs diligence on large data rooms will spend most of its credits on ingestion and Composer, not on the agent itself.

Blueflame AI pricing compared with Rogo, Hebbia and Model ML

Blueflame is the only one of the AI-native deal platforms that publishes any usage rates. The others sell seats on quotes, and the figures below are third-party estimates recorded on our price guides.

Blueflame's real advantage in that table is the data column. Rogo and Hebbia connect to PitchBook or LSEG only if you already hold those licenses, which can add $25,000 or more a seat. Blueflame sits on Grata's private company data and Datasite's data rooms, so a firm already on Grata buys less on top. Our guides on Rogo AI pricing per seat, Hebbia pricing and Model ML pricing go through each quote in detail, and the Finster AI pricing page covers the London-based rival that sells to the same banks.

Who owns Blueflame AI now?

Datasite does. Blueflame was founded in 2023 by Raj Bakhru and Henry Lindemann, raised a $5 million Series A in June 2024 at a reported $50 million valuation, and was acquired by Datasite on July 23, 2025, on undisclosed terms. Datasite had bought Grata seven weeks earlier, on June 3, 2025, and its controlling shareholder CapVest Partners committed $500 million to building out Datasite's intelligence products. Blueflame now calls itself "Datasite's AI agent for M&A" and runs inside Grata Intelligence.

That ownership shapes the price. You are no longer buying a standalone AI tool. You are buying into a deal stack that runs from sourcing in Grata to the virtual data room in Datasite, and the contract is likely to be bundled with whichever of those your firm already uses. If you are a Datasite VDR customer, ask whether Blueflame can be added to the existing agreement rather than priced fresh.

We found no Blueflame, Grata or Finster award in USAspending, so there is no public contract record to anchor the subscription layer the way a federal order did for BamSEC pricing.

Is Blueflame AI worth it?

It is worth pricing for a private equity firm, sell-side M&A team or private credit shop that already lives in Grata and Datasite, because the agent works on data you are already paying for, and the vendor reports large time savings on data room review. The usage model also rewards firms that keep jobs on the default agent rather than Composer.

It is a harder sell for a small sponsor or a search fund that mainly needs a quick view on public comparables and public acquirers. For that job a five-figure deal stack is the wrong tool. When the question is just whether a listed target or buyer is worth a closer look, a priced research card does it for far less. If you also need a rough number on a small private business before a first call, you can get an indicative valuation in seconds and save the deal platform for live processes.

What is a cheaper way to research public comps and acquirers?

Investables.ai writes a structured research card for any US ticker: what the business does, key market figures, comparables, the bull case, the bear case and labeled risks. The Analyst plan is $49 a month or $288 a year for 50 cards a month, and Team gives a deal team five seats for $249 a month. It does not search private companies, connect to a data room or build models, so it complements Blueflame rather than replacing it. For firms comparing the wider diligence category, our AI due diligence tool page sets out where a research card fits in the process.

Frequently asked questions about Blueflame AI pricing

Does Blueflame AI have a free trial?

Blueflame does not advertise a self-serve free trial. Access starts with a demo booked through blueflame.ai or grata.com, and firms usually negotiate a pilot for one team. Ask for the pilot's credit allowance, the credit price and what happens to ingested documents if you do not sign, since storage is billed per gigabyte per year.

Is Blueflame AI part of Grata?

Both are part of Datasite. Datasite acquired Grata on June 3, 2025 and Blueflame on July 23, 2025. Blueflame now runs inside Grata Intelligence, Grata's pricing page describes its data as connected to Blueflame, and Blueflame usage is billed on the Grata rate card, so in practice you buy them together.

How much is one Grata credit worth?

Grata does not publish a dollar value for a credit. The rate card lists what each action costs in credits, and the client's order form sets the price per credit and any committed volume. Ask sales for that figure in writing, along with overage terms, before you compare Blueflame with seat-priced tools like Rogo.