> ## Documentation Index
> Fetch the complete documentation index at: https://docs.paradime.io/llms.txt
> Use this file to discover all available pages before exploring further.
# Supported models
> Every model you can run DinoAI on and when to use each: Claude, Gemini, GPT-5.6, Grok, Kimi, and Qwen, from quick low-cost edits to deep refactors.
DinoAI runs on models from six providers. Pick one from the model picker in the DinoAI composer, in the editor or in an agent session, and the choice applies to that session. Models are grouped under their provider, and each bills at its own [DinoAI credit](/products/dino-ai/dinoai-credits) rate, so cheaper models stretch your credits further.
## Choosing a model
Not sure which to use? Stay on the default and DinoAI works well for almost everything. This table shows where to start for common analytics engineering tasks and when you might switch.
| Use case | Start with | Switch when |
| ------------------------------------------------------------------------ | -------------------------------- | ----------------------------------------------------------------------------------------------------------------------- |
| Everyday dbt™ development: writing and editing models, tests, YAML, docs | Sonnet 5 (default) | A hard multi-file refactor needs more depth (Opus 5), or a quick edit doesn't need the horsepower (a Flash-class model) |
| Quick questions and small edits | Gemini 3.7 Flash or GPT-5.6 Luna | The answer needs deeper project context or the change touches many files |
| Hard refactors, migrations, and stubborn debugging | Opus 5 or GPT-5.6 Sol | The heavy lifting is done and you're iterating on details |
| High-volume, cost-sensitive sessions | Qwen3 235B or Gemini 2.5 Flash | Accuracy matters more than cost |
| Methodical, step-by-step reasoning on a budget | Kimi K2 Thinking | You need faster answers |
## Available models
| Provider | Model | Best for |
| ----------- | ------------------------ | ------------------------------------------------------------------------------------------------------------------------------------------------------- |
| Anthropic | Sonnet 5 (default) | The all-rounder for analytics engineering: strong coding, strong tool use, sensible cost. Default for new sessions. |
| Anthropic | Sonnet 4.6 | Previous Sonnet generation with the same speed and capability balance. |
| Anthropic | Opus 5 | The deepest reasoning for gnarly refactors, migrations, and long agent runs. Premium-priced. Also runs the planning phase of agent sessions by default. |
| Anthropic | Opus 4.8 | Previous Opus generation for the hardest coding and reasoning work. |
| Anthropic | Haiku 4.5 | Fast, low-cost tasks. Runs subagent work inside sessions by default. |
| Google | Gemini 3.7 Flash | Latest Flash generation: fast coding, reasoning, and agentic workflows at low cost. |
| Google | Gemini 3.6 Flash | Fast coding and agentic workflows. |
| Google | Gemini 3.1 Pro (Preview) | Advanced coding, long-context understanding, and complex multi-step reasoning. Slower and premium-priced. |
| Google | Gemini 2.5 Pro | Deep reasoning and advanced coding; the most capable 2.5 model. |
| Google | Gemini 2.5 Flash | Everyday questions and edits balancing speed and intelligence. |
| xAI | Grok 4.20 | Reasoning at a mid-range rate, balancing capability and cost. |
| xAI | Grok 4.6 (Preview) | xAI's latest reasoning generation, in preview. |
| Moonshot AI | Kimi K2 Thinking | Step-by-step thinking at a low rate; methodical debugging and analysis on a budget. |
| Qwen | Qwen3 235B | The lowest-cost option for simple, high-volume tasks. |
| OpenAI | GPT-5.6 Sol | Flagship GPT-5.6: strongest for the hardest reasoning, coding, and agentic tasks. Premium-priced. |
| OpenAI | GPT-5.6 Terra | Balanced GPT-5.6 for everyday work at lower cost. |
| OpenAI | GPT-5.6 Luna | Fast, low-cost GPT-5.6 for simple or high-volume tasks. |
The lineup evolves as providers release and retire models; the model picker always shows the current list and the live credit rate for each model.
## Defaults
You don't have to pick a model to use DinoAI. Out of the box:
* **Sonnet 5** handles sessions where you haven't chosen a model
* **Haiku 4.5** runs subagent tasks spawned inside a session
* **Opus 5** runs the planning phase of agent sessions
## Related
* [DinoAI Credits](/products/dino-ai/dinoai-credits): how credits are calculated and Credit Saver Mode
* [Budget & Consumption](/products/dino-ai/budget-and-consumption): track spend and set monthly allowances
## Related topics
- [Gemini, GPT-5.6, Grok, Kimi, and Qwen come to DinoAI](/changelog/2026-08-26/multi-provider-models.md)
- [Model access](/guides/data-mesh-setup/model-access.md)
- [Column-Level Lineage Diff](/products/bolt/ci-cd/lineage-diff/index.md)
- [Asset Details](/products/data-catalog/asset-details/index.md)
- [Data Warehouse Tools](/products/dino-ai/tools-and-features/warehouse-tool/index.md)