Siesta AI
The right AI model for every task
Siesta AI routes every task to the right model and lets teams optimize for quality, cost, or balanced performance without changing the user experience.
See routing optionsChoose the outcome, not the model
Set routing policies for quality, cost, or balanced performance. Siesta AI evaluates each task and routes it to the most suitable model, while keeping governance, deployment, and regional controls aligned with your Azure environment.
Quality mode
Prioritize stronger reasoning models for complex analysis, coding, and high-stakes work.
Cost mode
Route routine prompts to faster, more efficient models to keep spend under control.
Balanced mode
Use a default policy that balances quality, latency, and cost for everyday work.
Model availability follows Microsoft Foundry Model Router support. Routing mode and selected model subset determine which models are eligible for each request.
All models in one place
Compare providers, regions, and model capabilities for your use case.
These models are eligible for Microsoft Foundry Model Router, so Siesta AI can route each request dynamically based on routing mode and selected model subset.
| Model | Regions | Context window | Best for |
|---|---|---|---|
| OpenAI | |||
GPT-4o
Siesta AI recommends for multimodal
|
|
128 000 | Multimodal |
|
GPT-4o Mini
|
|
128 000 | Speed & Efficiency |
|
GPT-4.1
|
|
1 047 576 | Coding |
|
GPT-4.1 Mini
|
|
1 047 576 | Speed & Efficiency |
|
GPT-4.1 Nano
|
|
1 047 576 | Speed & Efficiency |
|
o4-mini
|
|
200 000 | Reasoning |
|
GPT-5 Nano
|
|
400 000 | Speed & Efficiency |
|
GPT-5 Mini
|
|
400 000 | Speed & Efficiency |
|
GPT-5
|
|
400 000 | General Purpose |
|
GPT-5 Chat
|
|
128 000 | Conversations |
|
GPT-5.2
|
|
400 000 | Reasoning |
|
GPT-5.2 Chat
|
|
128 000 | Conversations |
|
GPT-5.3 Chat
Siesta AI recommends for conversations
|
|
128 000 | Conversations |
|
GPT-5.4 Nano
Siesta AI recommends for speed & efficiency
|
|
400 000 | Speed & Efficiency |
|
GPT-5.4 Mini
|
|
400 000 | Speed & Efficiency |
|
GPT-5.4
|
|
1 050 000 | General Purpose |
|
GPT-5.5
Siesta AI recommends for general purpose
|
|
1 050 000 | General Purpose |
|
GPT OSS 120B
Siesta AI recommends for enterprise workflows
|
|
131 072 | Enterprise Workflows |
| Anthropic | |||
Claude Haiku 4.5
|
|
200 000 | Speed & Efficiency |
|
Claude Sonnet 4.5
Siesta AI recommends for coding
|
|
200 000 | Coding |
|
Claude Opus 4.1
|
|
200 000 | Deep Research |
|
Claude Opus 4.6
|
|
200 000 | Reasoning |
|
Claude Opus 4.7
Siesta AI recommends for deep research
Siesta AI recommends for reasoning
|
|
200 000 | Deep Research |
| DeepSeek | |||
DeepSeek V3.1
|
|
128 000 | Data Analysis |
|
DeepSeek V3.2
Siesta AI recommends for data analysis
|
|
128 000 | Data Analysis |
| Meta | |||
Llama 4 Maverick
Siesta AI recommends for long context
|
|
1 000 000 | Long Context |
| xAI | |||
Grok 4
|
|
262 000 | Reasoning |
|
Grok 4 Fast Reasoning
Siesta AI recommends for reasoning
|
|
128 000 | Reasoning |
Router eligibility follows Microsoft Foundry Model Router support. Routing mode and selected subset determine which models can be used for each request.
Popular models Siesta AI can deploy or select directly through Azure AI Foundry when a team wants a specific model outside dynamic routing.
| Model | Regions | Context window | Best for |
|---|---|---|---|
Gemini 3.1 Pro
Siesta AI recommends for multimodal
|
|
1 000 000 | Multimodal |
|
Gemini 3 Flash
|
|
1 000 000 | Speed & Efficiency |
|
Gemini 3 Deep Think
|
|
1 000 000 | Reasoning |
| Mistral | |||
Mistral Large 3
Siesta AI recommends for general purpose
|
|
128 000 | General Purpose |
|
Mistral Medium 3.5
|
|
128 000 | General Purpose |
|
Codestral 2501
|
|
256 000 | Coding |
| Anthropic | |||
|
Claude Sonnet 5
Siesta AI recommends for coding
|
|
200 000 | Coding |
|
Claude Opus 4.8
|
|
200 000 | Reasoning |
|
Claude Sonnet 4.6
|
|
200 000 | General Purpose |
| DeepSeek | |||
|
DeepSeek V4 Pro
Siesta AI recommends for reasoning
|
|
128 000 | Reasoning |
|
DeepSeek V4 Flash
|
|
128 000 | Speed & Efficiency |
|
DeepSeek V3.2-Speciale
|
|
128 000 | Data Analysis |
| xAI | |||
|
Grok 4.3
|
|
262 000 | Reasoning |
|
Grok 4 20 Reasoning
|
|
128 000 | Reasoning |
|
Grok 4.1 Fast Non-Reasoning
|
|
128 000 | Speed & Efficiency |
| Meta | |||
|
Llama 4 Scout
Siesta AI recommends for long context
|
|
10 000 000 | Long Context |
|
Llama 3.3 70B
|
|
128 000 | General Purpose |
| Moonshot AI | |||
Kimi-K2.7-Code
|
|
256 000 | Coding |
|
Kimi-K2.6
|
|
256 000 | General Purpose |
|
Kimi-K2.5
|
|
256 000 | Long Context |
| Microsoft | |||
Phi-4
|
|
16 000 | Speed & Efficiency |
|
Phi-4-mini-instruct
|
|
128 000 | Speed & Efficiency |
| Cohere | |||
Cohere Command A
Siesta AI recommends for enterprise workflows
|
|
256 000 | Enterprise Workflows |
Direct models are curated catalog options. They can be deployed or selected directly, but they are not claimed as part of dynamic Model Router selection.
Start building with Siesta AI
See how Siesta AI can power your use case with the right model.
Book a demo