WEBVTT 1 00:00:00.630 --> 00:00:02.760 As you saw throughout this section, 2 00:00:02.760 --> 00:00:05.940 there's quite a lot you can do with open models 3 00:00:05.940 --> 00:00:07.260 through LM Studio, 4 00:00:07.260 --> 00:00:09.390 and there's quite a lot you can configure, 5 00:00:09.390 --> 00:00:11.220 but there's even more you can do 6 00:00:11.220 --> 00:00:14.490 with locally running models when using LM Studio, 7 00:00:14.490 --> 00:00:17.760 because you can not just use them through LM Studio, 8 00:00:17.760 --> 00:00:20.340 but you can also use them programmatically. 9 00:00:20.340 --> 00:00:22.890 So you could build your own applications 10 00:00:22.890 --> 00:00:26.820 that rely on your own locally running models, 11 00:00:26.820 --> 00:00:29.010 so that you don't have to pay for the usage 12 00:00:29.010 --> 00:00:32.940 of the API of some proprietary model out there. 13 00:00:32.940 --> 00:00:35.670 Now, just to make sure we're all on the same page, 14 00:00:35.670 --> 00:00:39.120 one of the most common ways of using large language models, 15 00:00:39.120 --> 00:00:41.220 no matter if we're talking about open, 16 00:00:41.220 --> 00:00:42.960 locally running models, 17 00:00:42.960 --> 00:00:44.790 as we covered it here in this course, 18 00:00:44.790 --> 00:00:47.430 or if we're talking about ChatGPT 19 00:00:47.430 --> 00:00:50.760 or Google Gemini or xAI's Grok, 20 00:00:50.760 --> 00:00:53.130 in many cases, when talking about 21 00:00:53.130 --> 00:00:55.080 such large language models, 22 00:00:55.080 --> 00:00:57.840 we're referring to those AI chatbots. 23 00:00:57.840 --> 00:00:59.730 And LM Studio is no exception. 24 00:00:59.730 --> 00:01:04.290 It's an AI chatbot just using our locally running model, 25 00:01:04.290 --> 00:01:07.980 but that's not the only way of using large language models. 26 00:01:07.980 --> 00:01:11.010 Another very powerful and popular way 27 00:01:11.010 --> 00:01:13.500 of using these large language models 28 00:01:13.500 --> 00:01:15.780 is in your own applications, 29 00:01:15.780 --> 00:01:16.950 your own tools, 30 00:01:16.950 --> 00:01:19.830 applications you might be selling to your customers, 31 00:01:19.830 --> 00:01:23.760 or any other kind of AI-powered application. 32 00:01:23.760 --> 00:01:26.190 Because, of course, you can use these models to, 33 00:01:26.190 --> 00:01:28.740 for example, parse user input, 34 00:01:28.740 --> 00:01:33.240 to build a program that automatically analyzes images, 35 00:01:33.240 --> 00:01:37.230 to build a program that automatically generates content. 36 00:01:37.230 --> 00:01:38.850 The sky is the limit, 37 00:01:38.850 --> 00:01:41.340 or your imagination is the limit. 38 00:01:41.340 --> 00:01:45.510 And you can do that with those proprietary models as well. 39 00:01:45.510 --> 00:01:49.350 For example, OpenAI also exposes their GPT models 40 00:01:49.350 --> 00:01:50.610 through an API, 41 00:01:50.610 --> 00:01:53.010 which you can consume programmatically, 42 00:01:53.010 --> 00:01:54.330 and you can learn more about that 43 00:01:54.330 --> 00:01:56.190 in their official documentation. 44 00:01:56.190 --> 00:01:58.290 That's not the focus of this course, 45 00:01:58.290 --> 00:02:00.840 but you can use these models programmatically. 46 00:02:00.840 --> 00:02:02.820 That's the important takeaway. 47 00:02:02.820 --> 00:02:06.240 And the same is true for those locally running models 48 00:02:06.240 --> 00:02:08.250 when using LM Studio, 49 00:02:08.250 --> 00:02:10.920 because LM Studio also exposes 50 00:02:10.920 --> 00:02:15.090 a locally running large language model API server, 51 00:02:15.090 --> 00:02:17.400 which you can access from inside your code 52 00:02:17.400 --> 00:02:20.910 to communicate with those locally running models. 53 00:02:20.910 --> 00:02:23.733 And that's exactly what we'll explore next.