Mistral, Qwen, DeepSeek and the Open Model Landscape
For technical leads and architects who need to understand the open-weight models beyond Llama, what each publisher offers, and how jurisdiction, license and hardware shape the right choice.
The curriculum is being prepared and reviewed. Enrollment is not open yet.
About this course
The strongest open-weight models no longer come from one company. European, Chinese and US labs now publish models that compete with closed systems on many tasks, under licenses ranging from Apache 2.0 and MIT to modified-MIT and custom community terms, and with very different implications for data jurisdiction depending on whether you use the publisher's hosted app or run the weights yourself. This course teaches you to read that landscape with the discipline of an architect rather than a leaderboard follower.
You will learn to read a model card (total versus active parameters, context length, modalities, training disclosures), then work through each major publisher: Mistral's Apache 2.0 models such as Mistral Large 3 and Mistral Small 4 alongside its premier and specialist models; the Qwen family's dense, mixture-of-experts, vision-language and coder variants and thinking modes; DeepSeek V4 and the reinforcement-learning reasoning approach popularized by R1; GLM and Kimi for long-context and agentic coding work; and Gemma, gpt-oss and Nemotron for deployment-sized models. The final modules cover matching model size to hardware, license fine print by publisher, government and procurement restrictions on certain models, and a worked selection for a hypothetical on-premises assistant. Facts are stated as of the review month and linked to official model cards. Completion and a passing exam score earn a Wirewalk Institute of AI certificate of completion and assessment; it is not endorsed by any vendor named.
What you will be able to do
Read an open-weight model card and extract parameters, active parameters, context, modalities, license and training disclosures
Compare the current model lines from Mistral AI, Qwen, DeepSeek, Z.ai and Moonshot AI and state what each is strongest at
Distinguish the data-jurisdiction implications of a publisher's hosted app or API from self-hosting the same weights
Select a deployment-sized open model such as Gemma, gpt-oss or Nemotron for a given hardware budget
Compare license terms across publishers and flag clauses that affect commercial use, redistribution or derivative models
Identify procurement, government and export-control constraints that may rule out a model in a given organization
Produce a documented open-model recommendation for a specific workload, hardware envelope and risk profile
Syllabus
6 modules · 17 lessons · about 12 hours
1
Module 1: Mapping the Open-Weight Field
Who publishes open weights, how to read their model cards, and what public rankings can and cannot tell you.3 lessons
Who Publishes Open Weights in 2026: European, Chinese and US Labs
Reading a Model Card: Total vs Active Parameters, Context, Modalities and Training Disclosures
Arenas and Open Leaderboards: Contamination, Self-Reported Scores and What Rankings Miss
2
Module 2: Mistral AI
Mistral's open and premier lines and its specialist models for speech, documents and code.3 lessons
Mistral's Apache 2.0 Line: Mistral Large 3, Mistral Small 4 and Ministral 3
Open vs Premier Models, Mistral AI Studio and Le Chat Enterprise
Voxtral, Mistral OCR and Codestral: Specialist Models for Speech, Documents and Code
3
Module 3: Qwen and DeepSeek
The two most widely deployed Chinese open-model families and the techniques behind them.3 lessons
The Qwen Family: Dense, MoE, Vision-Language and Coder Variants
Qwen Thinking Modes, Multilingual Strengths and Tool Calling
DeepSeek V4 and the R1 Lineage: MoE, Multi-Head Latent Attention and RL-Trained Reasoning
4
Module 4: Agentic Challengers and Deployment-Sized Models
GLM and Kimi for long-context agent work, and smaller open models that fit modest hardware.3 lessons
GLM and Kimi: Long-Context and Agentic Coding Contenders
Gemma, gpt-oss and Nemotron: Smaller Open Models Built for Deployment
Fitting a Shortlist to Hardware: Laptop GPU, Single Server or 8-GPU Node
5
Module 5: Jurisdiction, Licenses and Restrictions
The non-technical constraints that most often decide which open model you may use.3 lessons
Hosted Apps and APIs vs Self-Hosted Weights: Where Your Data Goes
License Fine Print by Publisher: Apache 2.0, MIT Variants and Custom Terms
Government Bans, Procurement Policies and Export Controls Affecting Model Choice
6
Module 6: Making the Call
Turning the landscape into a defensible recommendation for a real workload.2 lessons
Shortlisting by Workload: Coding, Multilingual Support, Document Extraction and Agents
Worked Selection: An Open Model for a Hypothetical On-Premises Contract-Review Assistant
Exam and certificate
The exam unlocks once you have completed every lesson. It draws 20 questions from a larger bank, spread across the modules, with the answer options shuffled for each attempt. You have 40 minutes and need 80% to pass. After a failed attempt you can try again after 24 hours, up to 3 attempts in total. Feedback is given per module, never as the answer key.
Passing earns a certificate of completion and assessment from the Wirewalk Institute of AI with a unique ID that anyone can check at institute.wirewalk.com/verify. It is not an accredited qualification or a vendor certification. How certification works.
The Wirewalk Institute of AI is independent. This course teaches publicly documented products and is not affiliated with or endorsed by Mistral AI, Alibaba (Qwen), DeepSeek, Z.ai (GLM), Moonshot AI (Kimi), Google (Gemma), OpenAI (gpt-oss), NVIDIA (Nemotron), or Hugging Face. Product and company names are trademarks of their respective owners.
A hands-on hour with ChatGPT, Claude and Gemini, told through one small business racing a deadline. Set up an account with a privacy setting you chose, decide what never goes into a personal account, write prompts that produce usable drafts, use files and web search, catch confident mistakes before they go out, and pick the assistant that fits how you work.
A practical, current tour of ChatGPT as a professional tool: choosing a plan and model, prompting well, analyzing files, researching with citations, using images and voice, organizing work in projects and memory, connecting apps and plugins, delegating multi-step work to ChatGPT Work, and knowing which admin and privacy controls matter. Ends with clear guidance on moving a workflow to the OpenAI API.
Learn Claude by solving one firm's real problem: getting through renewal season with too few analysts and without putting client data in the wrong place. Each module compares the approaches professionals actually choose between, shows a realistic failure and its fix, and ends with a Decision Lab where you make the call under pressure.