Models
AI-Starter supports AI models across multiple platforms. Below is the list of models available as official assistants.
Caution
Preview and experimental models may be discontinued without notice at the provider's discretion.
Model Characteristics
The following is based on evaluations within the development team. Each model has different strengths and characteristics.
| Series | Characteristics |
|---|---|
| Claude | Excels in text comprehension and coding capabilities with stable performance |
| Nova | Available at low cost |
| Gemini | Excellent cost-performance and multimodal processing |
| OpenAI | Rich tool ecosystem including web search and image generation |
AWS
Claude
| Model Name | Model ID | Description |
|---|---|---|
| Claude Haiku 5.5 | jp.anthropic.claude-haiku-5-5 | Lightweight and fast Claude model with reasoning support, available via Japan cross-region inference, with improved coding and agentic performance over Haiku 4.5 |
| Claude Sonnet 5.5 | global.anthropic.claude-sonnet-5-5 | High-performance model with reasoning support, improving on Sonnet 5 in coding and knowledge work at a lower cost per task |
| Claude Agent Global（複製推奨） | global.anthropic.claude-sonnet-5-5 | Sonnet 5.5-based template for tool-enabled agents; duplicate and configure it before use |
| Claude Agent JP（複製推奨） | jp.anthropic.claude-haiku-5-5 | Haiku 5.5-based template for tool-enabled agents, available via Japan cross-region inference; duplicate and configure it before use |
| Claude Opus 5.5 | jp.anthropic.claude-opus-5-5 | The most capable Claude model with reasoning support, available via Japan cross-region inference, with strong coding, knowledge work, and long-running task performance |
| Claude Haiku 5.5 (Messages API) | jp.anthropic.claude-haiku-5-5 | Calls Claude Haiku 5.5 through the Anthropic Messages API (bedrock-runtime InvokeModel). Does not support tool execution or spreadsheet input (see AnthropicMessages) |
Nova
| Model Name | Model ID | Description |
|---|---|---|
| Nova 2 Lite | jp.amazon.nova-2-lite-v1:0 | Supports file analysis with Code Interpreter |
Note
Models with the global. prefix are Global cross-Region inference profiles and are invoked through the bedrock-runtime endpoint (OpenAI-compatible API) in the region configured by bedrock_region.
Bedrock Grok
| Model Name | Model ID | Description |
|---|---|---|
| Bedrock Grok 4.6 | global.xai.grok-4.6 | High-performance model strong at coding and agentic tasks (reasoning effort low / medium / high / xhigh configurable) |
OpenAI Models via Bedrock
| Model Name | Model ID | Description |
|---|---|---|
| Bedrock GPT-6 Astra | global.openai.gpt-6-astra | Most capable reasoning model served via Bedrock (high cost) |
| Bedrock GPT-6.1 Sol | global.openai.gpt-6.1-sol | High-performance model served via Bedrock |
| Bedrock GPT-6.1 Terra | global.openai.gpt-6.1-terra | Balanced model served via Bedrock, combining performance and speed |
| Bedrock GPT-6 Luna | global.openai.gpt-6-luna | Fast model served via Bedrock |
| Bedrock GPT-6 Luna WebSearch | openai.gpt-6-luna | Fast model with web search (in-region model via bedrock-mantle; see Web Search) |
Caution
OpenAI models served through Bedrock handle data differently than other Bedrock models. Please review the notice in the assistant's description before enabling them.
OpenAI
| Model Name | Model ID | Description |
|---|---|---|
| GPT-6 Astra | gpt-6-astra | Most capable reasoning model with code execution, file analysis, and web search (high cost) |
| GPT-6.1 Sol | gpt-6.1-sol | High-performance model with code execution, file analysis, and web search |
| GPT-6.1 Terra | gpt-6.1-terra | Balanced model combining performance and speed, with code execution, file analysis, and web search |
| GPT-6 Luna | gpt-6-luna | Fast model with code execution, file analysis, and web search |
| GPT Image 2.5 Flare | gpt-image-2.5-flare | Fast, high-quality image generation model |
| GPT Image 2.5 Sunburst | gpt-image-2.5-sunburst | High-precision image generation and editing model |
Azure
| Model Name | Model ID | Description |
|---|---|---|
| Azure Model Router | model-router | Automatically selects the optimal model per request |
Google
| Model Name | Model ID | Description |
|---|---|---|
| Gemini 4 Argon | gemini-4-argon | High-performance model with reasoning and web search |
| Gemini 3.8 Flash | gemini-3.8-flash | Balanced model with reasoning and web search |
| Gemini 3.8 Flash Thinking | gemini-3.8-flash | Balanced model with enhanced reasoning and web search |
| Gemini 3.5 Flash-Lite | gemini-3.5-flash-lite | Fast GA model with reasoning and web search |
| Nano Banana 2.1 | gemini-nano-banana-2.1 | Generates both text and images, with improved image quality and text rendering over Nano Banana 2 |
| Nano Banana 2 Lite | gemini-3.1-flash-lite-image | Lightweight and fast image generation model |
Recommended Models
| Use Case | Recommended Model |
|---|---|
| General Q&A | Gemini 3.8 Flash / Claude Haiku 5.5 |
| Advanced Q&A | Claude Opus 5.5 / GPT-6 Astra / GPT-6.1 Sol / Gemini 4 Argon |
| Creative writing | Gemini 3.8 Flash |
| Coding | Claude Sonnet 5.5 / GPT-6.1 Sol |
| Business document assistance | Gemini 3.8 Flash |
| Translation assistance | Gemini 3.8 Flash |
| Long text comprehension | Gemini 3.8 Flash |
| Questions about images | Gemini 3.8 Flash |
| Questions about text files | Gemini 3.8 Flash |
| Questions about PDF files | Gemini 3.8 Flash |
| Questions about spreadsheets | GPT-6 Luna |
| Image generation | Nano Banana 2 Lite |
| Web search | GPT-6 Luna |
| Just want to try it out | Gemini 3.8 Flash / Claude Haiku 5.5 / GPT-6 Luna |
Additional Resources
The following external sites are helpful for comparing generative AI models. Note that information is regularly updated, so always verify the latest data and reliability.
Custom Assistants
In addition to the official assistants listed above, administrators can create custom assistants to use any model. See Assistants for details.