Free API as a basic offering, and charge for usage for advance information. Service offering can vary and there are many model to use. I think of it as a license to use my API

API as part of the product offering – which most successful companies use – and similar to YAMLI, build the basic function and allow the client to leverage it as needed.
The key factor, from my experience is to avoid becoming an API provider.

Now to the examples you gave za3tar:
- Ikbis could certainly benefit from having an API, it would help people build tools with which to post their photos and videos directly to it; services that provide digital photo printing could offer the possibility to pull photos from ikbis for users to print directly; and a bunch of other stuff as well. Just think of the YouTube API and what people have been doing with it as an example.

- When it comes to Maktoob and Jeeran, they’re not really one-service startups, they have many services and products, some of which could use an API and some that don’t need one. For example, Maktoob’s Kalimat Araby program will make good use of an API by letting agencies easily handle their sponsored link purchases; their As7ab service could also benefit from jumping on the whole social networking API bandwagon, opening their system to talk with other services like facebook, orkut, bebo and all.
- Dwwen might not really need an API in its current form, and that’s just fine, it’s not really a must.

YAMLI had to provide an API because that is the core of their business. Without their API they are nothing except a front for Google search, i.e. they are not providing any benefit.

So, i believe that startups that provide a service that is integratable with other services should provide an API .. otherwise, it is unnecessary.

Yamli on the other hand are doing a great job. The idea is brilliant and simple. The applications are really unlimited and the results are amazing. It’s a win-win-win situation.

Which leads me to the subject of my next point. Check your mail for additional details.

And that also kind of confirms my idea that if every service could just concentrate on building what they’re best at, developing their platform to do its job the best way possible, and optimizing their technical architecture to support their activities fully, then it could be easier for them to open it up later on.

In the case of Twitter launching the API turned the other way around as its the main reason why Twitter keeps on going down. Bottom line if your architecture is ready API is good, if not then No.