Deskpro AI supports a wide range of Large Language Models (LLMs) and Embedding Models across multiple providers, so you can choose the right option for your organization.
We’ve tested these models extensively to ensure compatibility for both prompting (LLM-powered tasks like summarization and reply suggestions) and embeddings (semantic search and data retrieval).
This article lists all currently supported public and private models, along with guidance for Cloud and Self-Hosted customers.
Prompting Model
A prompting model is like a translator for your request — it turns what you ask into clear, step-by-step instructions that the LLM can follow.
Example: You say, “Write a friendly email to reschedule my dentist appointment.” The prompting model reformats that request with tone and key points, and the LLM writes the email.
Embedding Model
Embeddings act like “meaning fingerprints” for text. An embedding model converts your words and documents into numeric vectors, allowing the LLM to find relevant information even when different words are used.
Example: You ask, “How do I get paid time off?” The embedding model retrieves the HR document that mentions “annual leave” and passes it to the LLM, so you get the correct policy in plain language.
Deskpro Cloud
Cloud customers can choose from a range of public AI providers and configure their account directly in Deskpro.
Provider | Vendor | Model | Model Name | Type |
|---|---|---|---|---|
Vertex | Gemini | gemini-2.0-flash-lite-001 | Prompting | |
Vertex | Gemini | gemini-embedding-001 | Embedding | |
OpenAI | OpenAI | Chat GPT4o mini | gpt-4o-mini | Prompting |
Open AI | OpenAI | Chat GPT4o mini | text-embedding-3-small | Embedding |
Azure | OpenAI | Chat GPT4o mini | gpt-4o-mini | Prompting |
Azure | OpenAI | Chat GPT4o mini | text-embedding-3-small | Embedding |
Deskpro | OpenAI | Chat GPT4o mini | gpt-4o-mini | Prompting |
Deskpro | OpenAI | Chat GPT4o mini | text-embedding-3-small | Embedding |
Anthropic | Anthropic | Claude 3.5 Haiku | gpt-4o-mini | Prompting |
Voyage | Voyage | Voyage 3.5 | voyage 3.5 | Embedding |
Mistral | Mistral | Mistral Medium 3.1 | magistral-medium-2507 | Prompting |
Mistral | Mistral | Mistral Medium 3.1 | mistral-embed | Embedding |
Deskpro Self-Hosted
With Deskpro Self-Hosted, you can either:
Subscribe to a public AI provider (any listed above), or;
Configure Deskpro to use their own self-hosted models through providers list below.
Provider | Vendor | Model | Model Name | Type |
|---|---|---|---|---|
Ollama | OpenAI | gpt-oss | Prompting | |
Ollama | Meta | llama 4 | llama4 | Prompting |
Ollama | Meta | llama 3.3 | llama3.3 | Prompting |
Ollama | Meta | llama 3.2 | llama3.2 | Prompting |
Ollama | Meta | llama 3.1 | llama3.1 | Prompting |
Ollama | Meta | llama 3 | llama3 | Prompting |
Ollama | Meta | llama 2 | llama2 | Prompting |
Ollama | Alibaba | qwen3 | qwen3 | Prompting |
Ollama | Alibaba | qwen2.5 | qwen2 | Prompting |
Ollama | Mistral | mistral-large | mistral-large-2411 | Prompting |
Ollama | Mistral | Mistral Nemo 12b | open-mistral-nemo-2407 | Prompting |
Ollama | Mistral | Mistral 24.11 | mistral | Prompting |
Ollama | Mistral | Mixtral 8x7B | mixtral | Prompting |
Ollama | Microsoft | Phi-4 | phi4 | Prompting |
Ollama | Microsoft | Phi-3.5 | phi3.5 | Prompting |
Ollama | Gemma 2 | gemma2 | Prompting | |
Ollama | Gemma 3n | gemma3n | Prompting | |
Ollama | Gemma 3n | gemma3 | Prompting | |
Ollama | Deepseek | Deepseek R1 | deepseek-r1 | Prompting |
Ollama | Nomic AI | gpt4all | nomic-embed-text | Embedding |
Ollama | Mixedbread | Mixedbread | mxbai-embed-large | Embedding |
Ollama | BAAI | BGE-M3 | bge-large | Embedding |
Ollama | Microsoft | MiniLM | all-minilm | Embedding |