> ## Documentation Index
> Fetch the complete documentation index at: https://docs.asteragents.com/llms.txt
> Use this file to discover all available pages before exploring further.
# Choosing a Model
> Pick the right LLM for your agent based on task complexity, speed, and cost
## Overview
Aster Agents supports **bring-your-own-API-key** across multiple providers. You connect your own API keys in **Control Hub > Providers**, then select a model when building each agent. This gives you full control over cost, performance, and provider choice.
**Don't use outdated models.** Models like GPT-4o, Claude 3.5 Sonnet, or GPT-4.1 are previous-generation and significantly underperform current models on agentic tasks. Always use the latest models listed below.
## Supported Providers
| Provider | API Key Location | Models |
| ---------------- | ------------------------------------------------------ | ----------------------------------------------------------------------------- |
| **Anthropic** | [console.anthropic.com](https://console.anthropic.com) | Claude Opus 4.7, Opus 4.6, Sonnet 4.6, Haiku 4.5 |
| **OpenAI** | [platform.openai.com](https://platform.openai.com) | GPT-5.6 (Sol / Terra / Luna), GPT-5.5, GPT-5.4 (+ Mini / Nano / Pro), GPT-4.1 |
| **Google** | [aistudio.google.dev](https://aistudio.google.dev) | Gemini 3.1 Pro, Gemini 3.5 Flash, Gemini 3 Flash |
| **xAI** | [console.x.ai](https://console.x.ai) | Grok 4-1 (+ fast reasoning / non-reasoning) |
| **Azure OpenAI** | Azure Portal | Any deployed model |
Add your API key in **Control Hub > Providers** to unlock that provider's models across all your agents.
## Model Identifiers
When configuring a model — in the UI, the API, or via the `manage_agents` tool — the platform expects the **`provider:model-id`** format, not the display name. `"claude-sonnet-4-6"` will not work; `"anthropic:claude-sonnet-4-6"` will.
| Display name | Model identifier |
| ----------------------------- | --------------------------------- |
| Claude Opus 4.7 | `anthropic:claude-opus-4-7` |
| Claude Opus 4.6 | `anthropic:claude-opus-4-6` |
| Claude Sonnet 5 | `anthropic:claude-sonnet-5` |
| Claude Sonnet 4.6 | `anthropic:claude-sonnet-4-6` |
| Claude Haiku 4.5 | `anthropic:claude-haiku-4-5` |
| GPT-5.6 Sol | `openai:gpt-5.6-sol` |
| GPT-5.6 Terra | `openai:gpt-5.6-terra` |
| GPT-5.6 Luna | `openai:gpt-5.6-luna` |
| GPT-5.5 | `openai:gpt-5.5` |
| GPT-5.4 | `openai:gpt-5.4` |
| GPT-5.4 Mini | `openai:gpt-5.4-mini` |
| GPT-5.4 Nano | `openai:gpt-5.4-nano` |
| GPT-5.4 Pro | `openai:gpt-5.4-pro` |
| GPT-4.1 | `openai:gpt-4.1` |
| Gemini 3.1 Pro | `google:gemini-3.1-pro-preview` |
| Gemini 3.5 Flash | `google:gemini-3.5-flash` |
| Gemini 3 Flash | `google:gemini-3-flash-preview` |
| Grok 4-1 | `xai:grok-4-1` |
| Grok 4-1 Fast (Reasoning) | `xai:grok-4-1-fast-reasoning` |
| Grok 4-1 Fast (Non-reasoning) | `xai:grok-4-1-fast-non-reasoning` |
| Azure deployment | `azure:` |
For the complete list including legacy and embedding models, see **Control Hub > Providers** in the dashboard.
## Recommended Models
These are the three models you should be choosing between for most agents.
The latest and greatest at mid-tier pricing — near-Opus quality on tool use, document analysis, structured output, and customer-facing work, but cheaper than Opus. The right default for most agents.
Reach for Opus when your agent regularly hits ambiguity, complex reasoning, or tasks where subtle errors have real consequences.
Pick GPT-5.4 when you need `web_search`, `code_interpreter`, or `image_generation`, or your team is already invested in the OpenAI ecosystem.
**Best all-around.** The default choice for most agents. Near-flagship intelligence at mid-tier pricing with the fastest output speed of any frontier model.
**Maximum capability.** The most capable model for tasks requiring deep reasoning, complex multi-step planning, or handling ambiguous instructions.
**Strong alternative.** OpenAI's flagship. Competitive with Claude Opus on overall benchmarks, with slightly different strengths.
| Spec | Claude Sonnet 4.6 | Claude Opus 4.6 | GPT-5.4 |
| -------------- | ------------------------------------------------------------------ | -------------------------------------------------------------------------- | ----------------------------------------------------------------------- |
| **Input** | \$3.00 / 1M tokens | \$5.00 / 1M tokens | \$2.50 / 1M tokens |
| **Output** | \$15.00 / 1M tokens | \$25.00 / 1M tokens | \$15.00 / 1M tokens |
| **Context** | 1M tokens | 1M tokens | 1M tokens |
| **Max Output** | 64K tokens | 128K tokens | 128K tokens |
| **Strengths** | Instruction following, tool use, structured output, long documents | Complex reasoning, nuanced judgment, long-form generation, ambiguous tasks | Broad knowledge, web search tool, image generation, code interpretation |
**Best for**: General-purpose agents, CRM workflows, document analysis, data extraction, customer-facing agents, scheduled tasks, multi-tool orchestration.
**Why Sonnet over Opus?** Sonnet 4.6 scores within 1-2% of Opus 4.6 on coding and agentic benchmarks (79.6% vs 80.8% on SWE-bench) while being significantly faster and 40% cheaper. For most agent tasks — querying databases, filling out forms, searching knowledge bases, writing emails — the difference is imperceptible.
**Best for**: Legal document analysis, financial modeling, research agents, complex multi-agent workflows, agents that need to handle edge cases gracefully, tasks where accuracy matters more than cost.
**When to choose Opus**: When your agent regularly encounters situations that require judgment calls — interpreting vague requirements, handling conflicting information, or producing output where subtle errors have real consequences.
**Best for**: Agents that need OpenAI's built-in tools (web search, code interpreter, image generation), teams already invested in the OpenAI ecosystem, general-purpose agents where input cost sensitivity matters.
**OpenAI-exclusive tools**: GPT-5.4 unlocks `web_search`, `code_interpreter`, and `image_generation` as provider tools — these aren't available with other providers.
## Budget-Friendly Options
For high-volume or cost-sensitive workloads:
| Model | Input | Output | Best For |
| -------------------- | ----------- | ----------- | ------------------------------------------------------- |
| **GPT-5.4 Mini** | \$0.75 / 1M | \$4.50 / 1M | Capable subagents, classification, extraction |
| **GPT-5.4 Nano** | \$0.20 / 1M | \$1.25 / 1M | High-volume routing, simple classification, triage |
| **Claude Haiku 4.5** | \$1.00 / 1M | \$5.00 / 1M | Fast responses, simple Q\&A, scheduled monitoring tasks |
These models are excellent for **multi-agent workflows** where a cheaper model handles routine subtasks (classification, data formatting, simple lookups) while a flagship model handles the complex reasoning.
## Choosing by Use Case
| Agent Type | Recommended Model | Why |
| ------------------------------ | ------------------------- | ------------------------------------------- |
| CRM data entry / updates | Sonnet 4.6 | Reliable tool use, fast, cost-effective |
| Document analysis & extraction | Sonnet 4.6 | Strong structured output, large context |
| Legal / compliance review | Opus 4.6 | Nuanced judgment, handles ambiguity |
| Financial analysis | Opus 4.6 | Complex reasoning, accuracy matters |
| Customer-facing chatbot | Sonnet 4.6 | Fast responses, good instruction following |
| Scheduled monitoring tasks | Haiku 4.5 | Runs frequently, cost matters, simple logic |
| Research agents | Opus 4.6 or GPT-5.4 | Deep reasoning + web search (GPT-5.4) |
| Multi-agent orchestrator | Sonnet 4.6 | Fast coordination, reliable tool dispatch |
| Subagent / worker | GPT-5.4 Mini or Haiku 4.5 | Cost-efficient for delegated subtasks |
| Data classification / triage | GPT-5.4 Nano | Ultra-cheap, high-volume |
## Cost Optimization Tips
1. **Start with Sonnet 4.6** — it handles 90%+ of agent tasks well. Only upgrade to Opus if you see quality issues on specific tasks.
2. **Use cheaper models for subagents** — in multi-agent workflows, the orchestrator can run on Sonnet while worker agents run on Haiku or GPT-5.4 Mini.
3. **Prompt caching saves money** — both Anthropic (90% off cached input) and OpenAI (50% off cached input) automatically cache repeated context. Agents with large system prompts or knowledge base content benefit significantly.
4. **Keep system prompts focused** — longer prompts cost more on every message. A concise, well-structured prompt outperforms a verbose one and costs less.
## Provider Tools
Some providers include built-in tools that extend your agent's capabilities beyond Aster's standard tool library:
| Provider | Tool | Description |
| ---------- | ---------------- | ---------------------------------------------- |
| **OpenAI** | Web Search | Search the web for current information |
| **OpenAI** | Code Interpreter | Write and execute Python code in a sandbox |
| **OpenAI** | Image Generation | Generate images from text prompts |
| **Google** | Google Search | Ground responses with real-time search results |
| **Google** | Code Execution | Generate and run Python code |
| **xAI** | X Search | Search X (Twitter) for posts and discussions |
| **xAI** | Web Search | Search the web for current information |
These tools appear automatically when you select a model from the corresponding provider. They can be enabled alongside Aster's standard tools.