AutoGen Beginner Course

Legacy code (2024). This course was recorded on AutoGen 0.2 (pyautogen==0.2.26) and the code is kept as it appears in the videos. Several models it uses have since been shut down, so those lessons fail with model_not_found until you change the model name. See Models used in this course were retired.

The AutoGen 0.2 line is now maintained as AG2 (pip install ag2); this course still pins the original pyautogen package.

• install pycharm community edition ide (if you don't have one)
• install anaconda (if you don't want to use venv)
• Reddit project is bonus, try it out!

Join my Discord community for any questions!

To install the requirements, type this in your terminal:
pip install -r requirements.txt

Use Python 3.11 or 3.12: pyautogen==0.2.26 requires Python below 3.13, and the pinned numpy==1.26.4 publishes wheels only up to 3.12.

Bonus Project Reddit URL: Create App

Models used in this course were retired

The substitutes are the ones listed on the OpenAI deprecations page:

Model in this course	Shut down	OpenAI's substitute
gpt-3.5-turbo	2026-10-23	gpt-5.6-terra
gpt-4	2026-10-23	gpt-5.6-sol
gpt-4-vision-preview	2024-12-06	a current vision model, e.g. gpt-5.6-terra
dall-e-3 (lesson 08)	2026-05-12	gpt-image-2.5-flare, called through the Images API (client.images.generate), not as a chat model in llm_config

gpt-5.4-mini is a cheaper option for these small demos ($0.75 / $4.50 per 1M input/output tokens, against $2 / $12 for gpt-5.6-terra and $4 / $20 for gpt-5.6-sol).

To run an affected lesson:

1. Change "model" in that lesson's OAI_CONFIG_LIST.json, and any filter_dict model list or "model": in its .py files (search the folder for gpt- and dall-e).
2. Delete "temperature" and "max_tokens" from the llm_config. GPT-5 models only accept the default temperature, and use max_completion_tokens instead of max_tokens.
3. Expect a cost-calculation warning from pyautogen 0.2: its price table predates these models. The chat still runs.

Lessons that reference a retired model: 01, 02, 03, 04, 05, 06, 07, 08, 10, 11, 13 and 14. Lessons 09 (LM Studio, local model) and 12 (Hugging Face Inference API) do not use OpenAI models.

For current, maintained AutoGen code (AutoGen 0.7 API), see tylerprogramming/ai.