What does Watson Cost? What is the Price?

8 thoughts on “What does Watson Cost? What is the Price?”

  1. Great article, esp. demystifying pricing, but apparently they moved their most valuable API discovery news, to Discovery, a separate app, could you plz clarify pricing– 0.10 per news query implies per web page? per pdf? per book? What size, and apparently this provides a digest? Also, price for RAM, is that weekly, monthly, one-time? What is enrichment? Thanks for answering my naive questions.

    1. Hi Rama, your question was well on time, since the pricing for Discovery Service was completely remade the 31st of July. It is now much simpler and more straight forward. Here is a link to the new pricing for Watson Discovery Service (incl Discovery News)

      I will try to anser your questions based on the new price model.

      1. Price per query.
      It is a price per API query. If you use the Tooling, it resembles one “query”, so it is not dependent of documents, pdfs or similar. As an example. If you have a chatbot that helps you search for news through Discovery News, one query is when you ask Discovery for news e.g “positive news about Electrolux”.

      In the new price-model you can ask 1000 queries for free and thereafter it is $0.10 per query as you wrote.

      2. If you want to use Discovery Service, and not the already pre-enriched Discovery News, to ingest you own data, then there is a volume-based pricing as well. Today WDS supports, html, pdf, word and JSON. If I am not mistaken pdfs and word docs are converted to html and then to JSON by WDS.

      The current limits are as follows: 2000 docs for free (200mb in max 2 collections). See picture below for additional tiers.

      3. Prices for RAM. Those are also updated, but now in a simpler way. See image on new pricing for full description. They are all based on a monthly period for invoicing (even though invoicing usually lags a bit).

      4. What is enrichments? That is the main purpose of using Discovery Service (or Discovery News), it adds cognitive information to unstructured data. One way it does this in Discovery News is that it determines if a news has a positive or negative sentiment (including a confidence score), other enrichments are entities, categories etc which results in Discovery News to be able to tell us that a news is about Champions League, Real Madrid and Christiano Ronaldo etc. As of recently it is the same as Natural Language Understanding, but stored in the cloud instead of just passing the enrichments through an API as NLU does. You can also create your own models using Watson Knowledge Studio (and create your own enrichments that apply to your specific domain, as IBM has done with News, that model is specifically built for news).

      No question is naive in regards to Watson and AI etc, it is a new field and it is not the easiest one to digest. I hope this made some sense to you, otherwise just reply.

Leave a Reply

Your email address will not be published. Required fields are marked *