tota supports multiple AI providers. You can configure one or more providers, select a default model, and use automatic fallback when the primary fails.

#Quick setup (recommended)

1. Create an API key from your provider (links below).
2. Run the interactive setup:
   Copy

   tota setup llm
   

3. Paste the key when prompted and pick a model from the fetched list.
4. Repeat for any extra providers you want in the fallback chain.
5. Verify your configuration:
   Copy

   tota status
   

If you need to change providers later, rerun tota setup llm.

#Supported providers and key links

#Provider signup steps (official links)

Most pages require login. If a link redirects, sign in and look for API keys or Developer settings in the account menu.

#OpenAI

1. Open https://platform.openai.com/api-keys and sign in.
2. Click Create API Key.
3. Copy the key and store it safely.

#Anthropic

1. Open https://console.anthropic.com/settings/keys (redirects to Claude Console).
2. Click Create key and copy it.

#DeepSeek

1. Open https://platform.deepseek.com/api_keys and sign in.
2. Create a new key and copy it.

#Grok (xAI)

1. Open https://console.x.ai/ and sign in.
2. Go to API keys and generate a key.

#Groq

1. Open https://console.groq.com/keys and sign in.
2. Click Create API Key, copy the key (starts with gsk_).

#Ollama Cloud

1. Open https://ollama.com/settings and sign in.
2. Create an API key under API Keys.

#Ollama Local

1. Install Ollama from https://ollama.com/download.
2. Pull a model (example below).

#NVIDIA NIM

1. Open https://build.nvidia.com/settings/api-keys and sign in.
2. Create an API key in the settings page.

#OpenRouter

1. Open https://openrouter.ai/keys and sign in.
2. Create a key in your workspace.

#MiMo (Xiaomi)

1. Sign in to the MiMo developer portal (start at https://api.xiaomimimo.com).
2. Create an API key in your account settings.

#Manual configuration (optional)

If you prefer not to use the wizard, add keys and settings to ~/.tota/.env and restart tota. Model names are examples; choose from your provider's model list.

# OpenAI
OPENAI_API_KEY=sk-your-key
OPENAI_MODEL=gpt-4o-mini

# Anthropic
ANTHROPIC_API_KEY=your-key
ANTHROPIC_MODEL=claude-sonnet-4-20250514

# DeepSeek
DEEPSEEK_API_KEY=sk-your-key
DEEPSEEK_BASE_URL=https://api.deepseek.com/v1
DEEPSEEK_MODEL=deepseek-chat

# Grok (xAI)
GROK_API_KEY=your-key
GROK_BASE_URL=https://api.x.ai/v1
GROK_MODEL=grok-4

# Groq
GROQ_API_KEY=gsk_your-key
GROQ_BASE_URL=https://api.groq.com/openai/v1
GROQ_MODEL=llama-3.3-70b-versatile

# Ollama Cloud
OLLAMA_CLOUD_API_KEY=your-key
OLLAMA_CLOUD_BASE_URL=https://ollama.com/v1
OLLAMA_CLOUD_MODEL=gpt-oss:120b

# Ollama Local (no key required)
OLLAMA_LOCAL_BASE_URL=http://127.0.0.1:11434/v1
OLLAMA_LOCAL_MODEL=qwen3.5:2b
OLLAMA_LOCAL_ENABLED=true

# OpenAI-compatible (custom)
OPENAI_COMPAT_API_KEY=your-key
OPENAI_COMPAT_BASE_URL=https://your-openai-compatible-host/v1
OPENAI_COMPAT_MODEL=your-model
OPENAI_COMPAT_ENABLED=true

# NVIDIA NIM
NVIDIA_API_KEY=nvapi-your-key
NVIDIA_BASE_URL=https://integrate.api.nvidia.com/v1
NVIDIA_MODEL=nvidia/nemotron-3-super-120b-a12b

# OpenRouter
OPENROUTER_API_KEY=sk-or-your-key
OPENROUTER_BASE_URL=https://openrouter.ai/api/v1
OPENROUTER_MODEL=openrouter/auto

# MiMo (Xiaomi)
MIMO_API_KEY=sk-your-key
MIMO_BASE_URL=https://api.xiaomimimo.com/v1
MIMO_MODEL=mimo-v2.5-pro

# MiMo Token Plan
MIMO_TOKEN_PLAN_API_KEY=tp-your-key
MIMO_TOKEN_PLAN_BASE_URL=https://token-plan-cn.xiaomimimo.com/v1
MIMO_TOKEN_PLAN_MODEL=mimo-v2.5-pro
MIMO_TOKEN_PLAN_ENABLED=true

# Default provider (optional)
DEFAULT_PROVIDER=deepseek

#Provider-specific notes

#OpenAI-compatible (custom)

Use this when you have a self-hosted or third-party OpenAI-compatible endpoint.

1. Set OPENAI_COMPAT_BASE_URL to the provider base URL.
2. Set OPENAI_COMPAT_MODEL to the model name they expose.
3. Add OPENAI_COMPAT_API_KEY if the endpoint requires it.

#Ollama Local

1. Install Ollama from https://ollama.com/download.
   • macOS: download the DMG from the site.
   • Linux: curl -fsSL https://ollama.com/install.sh | sh
2. Pull a model:
   Copy

   ollama pull qwen3.5:2b
   

3. Keep OLLAMA_LOCAL_BASE_URL pointed at http://127.0.0.1:11434/v1 (or /api).

#Ollama Cloud

Ollama Cloud uses the OpenAI-compatible endpoint at https://ollama.com/v1 and requires OLLAMA_CLOUD_API_KEY.

#Key format checks

tota validates API key formats before accepting them:

• OpenAI: sk-, sk-proj-, or sk-svcacct-
• Anthropic: sk-ant-
• DeepSeek: sk-
• MiMo: sk-
• MiMo Token Plan: tp-
• NVIDIA NIM: nvapi-
• OpenRouter: sk-or-
• Grok and Ollama Cloud: long token, no spaces
• Groq: gsk_

#Common issues