Config & Models Commands

Configuration commands manage system settings, API keys, and LLM model selection in AgentOS.

Config Commands

config show

Display the current configuration.

agentos config show

Example output:

default_model = "claude-opus-4-6"
api_url = "http://localhost:3111"

[keys]
anthropic = "sk-ant-..."
openai = "sk-..."

config get

Get the value of a specific configuration key.

agentos config get <key>

key

string

required

Configuration key to retrieve

Example:

agentos config get default_model

Output:

default_model = claude-opus-4-6

config set

Set a configuration value.

agentos config set <key> <value>

key

string

required

Configuration key to set

value

string

required

Value to set

Example:

agentos config set default_model gemini-2.0-flash

Output:

✓ Set default_model = gemini-2.0-flash

config unset

Remove a configuration key.

agentos config unset <key>

key

string

required

Configuration key to remove

Example:

agentos config unset custom_setting

Output:

✓ Removed key: custom_setting

config set-key

Set an API key for an LLM provider.

agentos config set-key <provider> <key>

provider

string

required

Provider name (e.g., anthropic, openai, google)

key

string

required

API key value

Example:

agentos config set-key anthropic $ANTHROPIC_API_KEY
agentos config set-key openai sk-proj-...
agentos config set-key google $GOOGLE_API_KEY

Output:

✓ API key set for anthropic

API keys are stored in ~/.agentos/config.toml under the [keys] section.

config keys

List all configured API keys (masked for security).

agentos config keys

Example output:

PROVIDER             STATUS
anthropic            sk-a...xyz
openai              sk-p...abc
google              AIza...def
groq                gsk-...ghi

Models Commands

Manage and explore available LLM models.

models list

List all available models with pricing and capabilities.

agentos models list

Example output:

MODEL                     PROVIDER        TIER         CONTEXT    PRICE (in/out)
claude-opus-4-6           anthropic       premium      200K       $15/$75
claude-sonnet-4-6         anthropic       standard     200K       $3/$15
claude-haiku-3-5          anthropic       fast         200K       $0.80/$4
gpt-4o                    openai          premium      128K       $2.50/$10
gpt-4o-mini               openai          fast         128K       $0.15/$0.60
o1                        openai          reasoning    200K       $15/$60
o3-mini                   openai          reasoning    200K       $1.10/$4.40
gemini-2.0-flash          google          fast         1000K      $0.075/$0.30
gemini-2.0-pro            google          standard     2000K      $1.25/$5
deepseek-v3               deepseek        standard     64K        $0.27/$1.10
deepseek-r1               deepseek        reasoning    64K        $0.55/$2.19
llama-3.3-70b             groq            standard     128K       $0.59/$0.79
mistral-large             mistral         premium      128K       $2/$6
command-r-plus            cohere          standard     128K       $2.50/$10

Columns explained:

TIER: fast (quick/cheap), standard (balanced), premium (best quality), reasoning (o1/o3 style)
CONTEXT: Context window size in tokens (K = 1,000)
PRICE: Cost per million tokens (input/output) in USD

models aliases

List model aliases and their mappings.

agentos models aliases

Example output:

  opus → claude-opus-4-6
  sonnet → claude-sonnet-4-6
  haiku → claude-haiku-3-5
  gpt4 → gpt-4o
  gpt4-mini → gpt-4o-mini
  gemini → gemini-2.0-flash
  fast → gemini-2.0-flash
  premium → claude-opus-4-6
  reasoning → o1

Use aliases in configuration:

agentos config set default_model opus

models providers

List all LLM providers and their status.

agentos models providers

Example output:

  ● Anthropic          (3 models)
  ● OpenAI            (4 models)
  ● Google            (2 models)
  ● DeepSeek          (2 models)
  ● Groq              (5 models)
  ● Mistral           (3 models)
  ● Cohere            (2 models)
  ○ AWS Bedrock       (0 models)
  ○ Azure OpenAI      (0 models)

● Green dot = Provider configured with API key
○ Red dot = Provider not configured

models describe

Get detailed information about a specific model.

agentos models describe <model>

model

string

required

Model ID or alias

Example:

agentos models describe claude-opus-4-6

Output:

{
  "id": "claude-opus-4-6",
  "provider": "anthropic",
  "name": "Claude Opus 4.6",
  "tier": "premium",
  "contextWindow": 200000,
  "maxOutputTokens": 16384,
  "inputPrice": 15.0,
  "outputPrice": 75.0,
  "capabilities": [
    "chat",
    "function-calling",
    "vision",
    "streaming"
  ],
  "strengths": [
    "Complex reasoning",
    "Code generation",
    "Long context understanding",
    "Multilingual"
  ],
  "released": "2026-02-15"
}

Supported LLM Providers

AgentOS supports 25 LLM providers:

Major Providers
Fast Inference
Open Source
Specialized
Routing

Anthropic - Claude Opus, Sonnet, Haiku
OpenAI - GPT-4o, GPT-4o-mini, o1, o3-mini
Google - Gemini 2.0 Flash, Pro
AWS Bedrock - Claude, Titan
Azure OpenAI - GPT models on Azure

Model Selection

AgentOS uses complexity-based model selection: Complexity scoring factors:

Input token count
Task type (code, reasoning, chat)
Context requirements
Budget constraints

Override with explicit model selection:

agentos config set default_model claude-opus-4-6

Configuration File Structure

Location: ~/.agentos/config.toml

# Global settings
default_model = "claude-opus-4-6"
api_url = "http://localhost:3111"
log_level = "info"

# API keys
[keys]
anthropic = "sk-ant-..."
openai = "sk-proj-..."
google = "AIza..."
groq = "gsk-..."

# Model preferences
[models]
prefer_fast = ["gemini-2.0-flash", "gpt-4o-mini"]
prefer_reasoning = ["o1", "deepseek-r1"]
prefer_premium = ["claude-opus-4-6", "gpt-4o"]

# Budget limits
[budget]
max_tokens_per_day = 1000000
max_cost_per_day = 100.0

# Security
[security]
audit_enabled = true
approval_tier = "async"

Environment Variables

You can also configure via environment variables:

# API keys
export ANTHROPIC_API_KEY="sk-ant-..."
export OPENAI_API_KEY="sk-proj-..."
export GOOGLE_API_KEY="AIza..."

# Settings
export AGENTOS_API_URL="http://localhost:3111"
export AGENTOS_DEFAULT_MODEL="claude-opus-4-6"
export AGENTOS_LOG_LEVEL="debug"

Priority order:

Command-line arguments
Environment variables
Config file (~/.agentos/config.toml)
Defaults

Examples

# Initialize AgentOS
agentos init --quick

# Set API keys
agentos config set-key anthropic $ANTHROPIC_API_KEY
agentos config set-key openai $OPENAI_API_KEY

# Set default model
agentos config set default_model opus

# Verify configuration
agentos config show

Model Comparison

Use Case	Recommended Model	Why
Quick chat	gemini-2.0-flash, gpt-4o-mini	Fast, cheap, good quality
Code generation	claude-sonnet-4-6, gpt-4o	Best code understanding
Complex reasoning	claude-opus-4-6, o1	Superior reasoning ability
Long context	gemini-2.0-pro	2M token context window
Budget-conscious	haiku, gpt-4o-mini	Lowest cost per token
Math/logic	o1, deepseek-r1	Specialized reasoning
Local/offline	ollama (llama-3.3-70b)	No API needed
Ultra-fast	groq (llama-3.3-70b)	Fastest inference

Best Practices

Secure API keys

Never commit API keys to version control. Use environment variables or the config file (which should be gitignored).

Use aliases

Use model aliases like opus or fast instead of full model names for easier switching.

Set budget limits

Configure daily token and cost limits to prevent unexpected charges.

Multiple providers

Configure multiple providers for fallback and cost optimization.

Model-specific agents

Create agents with specific models for specialized tasks (e.g., reasoning agent with o1).

Next Steps

Agent Commands

Create agents with specific models

Security

Configure security and budget limits

Workflows

Use different models in workflows

CLI Overview

Back to CLI overview

Get Started

Core Concepts

CLI Reference

Tools & Capabilities

Control Plane

Security

Advanced Features

Templates & Examples

Config & Models Commands

Config Commands

config show

config get

config set

config unset

config set-key

config keys

Models Commands

models list

models aliases

models providers

models describe

Supported LLM Providers

Model Selection

Configuration File Structure

Environment Variables

Examples

Model Comparison

Best Practices

Next Steps

Agent Commands

Security

Workflows

CLI Overview

Build docs developers (and LLMs) love

Get Started

Core Concepts

CLI Reference

Tools & Capabilities

Control Plane

Security

Advanced Features

Templates & Examples

​Config Commands

​config show

​config get

​config set

​config unset

​config set-key

​config keys

​Models Commands

​models list

​models aliases

​models providers

​models describe

​Supported LLM Providers

​Model Selection

​Configuration File Structure

​Environment Variables

​Examples

​Model Comparison

​Best Practices

​Next Steps

Agent Commands

Security

Workflows

CLI Overview

Build docs developers (and LLMs) love

Config Commands

config show

config get

config set

config unset

config set-key

config keys

Models Commands

models list

models aliases

models providers

models describe

Supported LLM Providers

Model Selection

Configuration File Structure

Environment Variables

Examples

Model Comparison

Best Practices

Next Steps