# Dual Language Chatbot **URL:** \ **Category:** Rasa Open Source\ **Created:** [October 23, 2022, 1:41pm UTC](https://forum.rasa.com/t/dual-language-chatbot/55899 "2022-10-23T13:41:36Z")\ **Posts on this page:** 6\ **Page:** 1
**Author:** ![Abo3bah](https://avatars.discourse-cdn.com/v4/letter/a/5f8ce5/32.png) [@Abo3bah](https://forum.rasa.com/u/Abo3bah)\ **Post date:** [October 23, 2022, 1:41pm UTC](https://forum.rasa.com/t/dual-language-chatbot/55899/1 "2022-10-23T13:41:36Z")
hello! I was working with [botfront.io](https://botfront.io/) with `rasa 2.3` but we was facing some issues due to lower version of rasa that’s why we take the decision to work with `rasa x 1.1.0 with rasa 3.0.8` I use to implement dual language chatbots mainly (English - Arabic), `Botfront` was helping us by providing `separate NLU files` for each language `with its own Configuration` “as shown in the images” which was giving me the availability to tune each language in a separate way which was having a great effect on the bot performance so does anyone know how can I do that in rasa x with the versions `rasa x 1.1.0 and rasa 3.0.8`? Or even tell me what is the best practices for rasa in building a `dual language bot`? **the target is to have separate NLU files with separate configuration for each one of them for the same model like `botfront` was doing** I attached some images to show you how `Botfront` was dealing with the dual language ![Ar_nlu_ed](https://europe1.discourse-cdn.com/flex013/uploads/rasa/original/3X/6/9/69da1c95b61a97fd4efb987984060b82ce290668.png) ![Ar_settings_ed](https://europe1.discourse-cdn.com/flex013/uploads/rasa/original/3X/1/4/14a3a4aab78e456bad5807cf3217f16141574a28.png) ![En_nlu_ed](https://europe1.discourse-cdn.com/flex013/uploads/rasa/original/3X/c/3/c338378feca6e0a1cfa608b6e3a50112bad7d2f0.png) ![En_settings_ed](https://europe1.discourse-cdn.com/flex013/uploads/rasa/original/3X/1/d/1de63301e48d0deecbd50edfdf8d531f5c84fe2e.png) ---
**Author:** ![AhmedMagdy-96](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/ahmedmagdy-96/32/21583_2.png) [@AhmedMagdy-96](https://forum.rasa.com/u/AhmedMagdy-96)\ **Post date:** [October 23, 2022, 1:52pm UTC](https://forum.rasa.com/t/dual-language-chatbot/55899/2 "2022-10-23T13:52:23Z")
I have a similar issue with building a multi-language chatbot with two separate NLU and Config files on RasaX version 1.1.0 ---
**Author:** ![Abo3bah](https://avatars.discourse-cdn.com/v4/letter/a/5f8ce5/32.png) [@Abo3bah](https://forum.rasa.com/u/Abo3bah)\ **Post date:** [October 27, 2022, 2:16pm UTC](https://forum.rasa.com/t/dual-language-chatbot/55899/3 "2022-10-27T14:16:28Z")
Anyone can help please?! ---
**Author:** ![rasa\_learner](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/rasa_learner/32/21721_2.png) [@rasa\_learner](https://forum.rasa.com/u/rasa_learner)\ **Post date:** [October 27, 2022, 3:09pm UTC](https://forum.rasa.com/t/dual-language-chatbot/55899/4 "2022-10-27T15:09:39Z")
Hi @Abo3bah, you can build individual models (either NLU only or NLU + Core) for respective languages and serve them at different ports on the same server. You will also be able to utilize a single action server for both of the models. ---
**Author:** ![Abo3bah](https://avatars.discourse-cdn.com/v4/letter/a/5f8ce5/32.png) [@Abo3bah](https://forum.rasa.com/u/Abo3bah)\ **Post date:** [November 1, 2022, 8:09am UTC](https://forum.rasa.com/t/dual-language-chatbot/55899/5 "2022-11-01T08:09:09Z")
Hi @rasa_learner, I know that but I do not want to have two separate instances at different ports ---
**Author:** ![rasa\_learner](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/rasa_learner/32/21721_2.png) [@rasa\_learner](https://forum.rasa.com/u/rasa_learner)\ **Post date:** [November 2, 2022, 1:54pm UTC](https://forum.rasa.com/t/dual-language-chatbot/55899/6 "2022-11-02T13:54:56Z")
Hi @Abo3bah, it would be the fitting approach to serve multiple NLU models - [Using NLU Only](https://rasa.com/docs/rasa/nlu-only/#training-nlu-only-models:~:text=Version%3A%203.x-,Using%20NLU%20Only,-Find%20out%20how)