AI Models
The underlying AI models that power our agents, each with different capabilities.
Auto
Automatic model routing through the Vercel AI Gateway. Picks the best available model for each request and falls back to alternates if the primary is overloaded or unavailable.
Automatically picks the best available model with seamless fallback if any provider is overloaded.
Anthropic Claude
Anthropic is an AI safety company and public benefit corporation that builds the Claude family of large language models. Founded in 2021, Anthropic focuses on developing reliable, steerable AI designed to be helpful, harmless, and honest.
Fast and intelligent model for quick tasks
The best combination of speed and intelligence, with adaptive thinking and long context support
Previous-generation Sonnet with long context and extended thinking
Most capable Claude model with a step-change in agentic coding and complex reasoning
Previous-generation Opus with a step-change in agentic coding and complex reasoning
Previous-generation Opus with adaptive thinking for complex reasoning and agentic workflows
Google Gemini
Google DeepMind's Gemini models offer multimodal understanding with large context windows, strong reasoning, and efficient performance across tasks.
Near-Pro reasoning and coding at Flash-tier cost and speed. Dynamic thinking is on by default.
Google's most advanced Gemini 3.1 model with powerful agentic capabilities, multimodal understanding, and state-of-the-art reasoning.
Mistral AI
Mistral AI builds efficient, open-weight language models. Known for strong multilingual support, fast inference, and competitive performance at lower cost.
Hybrid model optimized for general chat, coding, agentic tasks, and complex reasoning with text and image input support
Balanced model for most tasks with good performance and cost efficiency
Mistral's most capable model with strong multilingual support and advanced reasoning for complex tasks
OpenAI GPT
OpenAI's GPT models deliver strong general-purpose intelligence with broad knowledge, creative writing, and code generation capabilities.
OpenAI's frontier model — fully-retrained base since GPT-4.5, with 1M context window and strong general reasoning. Priced like Claude Opus; available on the Pro plan and above.
GPT-5.4 with a 1M context window, built-in computer-use capabilities, and optional reasoning. Sonnet-equivalent OpenAI option for Basic+ users.
GPT 5.4 mini is an optimized version of GPT 5.4 — fast and efficient for everyday tasks.
Last version of GPT 4
OpenAI's open-weights 120B-parameter model.
xAI
xAI's Grok models — frontier reasoning with very large context windows.
xAI's fast model with an unprecedented 2 million token context window.
xAI's advanced reasoning model with strong coding and agentic capabilities and a 1 million token context window.
DeepSeek
DeepSeek's efficient open models with strong coding and math reasoning.
DeepSeek's efficient reasoning model with strong coding and math capabilities.
DeepSeek's fast and efficient V4 model optimized for speed with a 1 million token context window.
DeepSeek's flagship V4 reasoning model with strong coding and math capabilities and a 1 million token context window.
Moonshot AI
Moonshot AI's Kimi models — long-context reasoning.
Moonshot AI's advanced reasoning model with strong performance on complex tasks.
Z-AI
Z-AI's GLM models — fast, multilingual general-purpose intelligence.
Z-AI's fast and efficient general-purpose model with strong multilingual capabilities.
MiniMax
MiniMax's general-purpose models with large context support.
MiniMax's latest general-purpose model with strong reasoning and large context support.
Qwen
Alibaba's Qwen models — high-performance reasoning with up to 1M-token context.
Qwen's flagship model with state-of-the-art reasoning and a 1 million token context window.
Qwen's high-performance model with 1 million token context window.
Qwen's fast and efficient model optimized for speed with 1M context.
Auto
Automatic model routing through the Vercel AI Gateway. Picks the best available model for each request and falls back to alternates if the primary is overloaded or unavailable.
Automatically picks the best available model with seamless fallback if any provider is overloaded.
auto Anthropic Claude
Anthropic is an AI safety company and public benefit corporation that builds the Claude family of large language models. Founded in 2021, Anthropic focuses on developing reliable, steerable AI designed to be helpful, harmless, and honest.
Fast and intelligent model for quick tasks
claude-haiku-4-5 The best combination of speed and intelligence, with adaptive thinking and long context support
claude-sonnet-4-6 Previous-generation Sonnet with long context and extended thinking
claude-sonnet-4-5 Most capable Claude model with a step-change in agentic coding and complex reasoning
claude-opus-4-8 Previous-generation Opus with a step-change in agentic coding and complex reasoning
claude-opus-4-7 Previous-generation Opus with adaptive thinking for complex reasoning and agentic workflows
claude-opus-4-6 Google Gemini
Google DeepMind's Gemini models offer multimodal understanding with large context windows, strong reasoning, and efficient performance across tasks.
Near-Pro reasoning and coding at Flash-tier cost and speed. Dynamic thinking is on by default.
google/gemini-3.5-flash Google's most advanced Gemini 3.1 model with powerful agentic capabilities, multimodal understanding, and state-of-the-art reasoning.
google/gemini-3.1-pro-preview Mistral AI
Mistral AI builds efficient, open-weight language models. Known for strong multilingual support, fast inference, and competitive performance at lower cost.
Hybrid model optimized for general chat, coding, agentic tasks, and complex reasoning with text and image input support
mistral-small-latest Balanced model for most tasks with good performance and cost efficiency
mistral-medium-latest Mistral's most capable model with strong multilingual support and advanced reasoning for complex tasks
mistral-large-latest OpenAI GPT
OpenAI's GPT models deliver strong general-purpose intelligence with broad knowledge, creative writing, and code generation capabilities.
OpenAI's frontier model — fully-retrained base since GPT-4.5, with 1M context window and strong general reasoning. Priced like Claude Opus; available on the Pro plan and above.
gpt-5.5 GPT-5.4 with a 1M context window, built-in computer-use capabilities, and optional reasoning. Sonnet-equivalent OpenAI option for Basic+ users.
gpt-5.4 GPT 5.4 mini is an optimized version of GPT 5.4 — fast and efficient for everyday tasks.
gpt-5.4-mini Last version of GPT 4
gpt-4.1 OpenAI's open-weights 120B-parameter model.
gpt-oss-120b xAI
xAI's Grok models — frontier reasoning with very large context windows.
xAI's fast model with an unprecedented 2 million token context window.
xai/grok-4.1-fast xAI's advanced reasoning model with strong coding and agentic capabilities and a 1 million token context window.
xai/grok-4.3 DeepSeek
DeepSeek's efficient open models with strong coding and math reasoning.
DeepSeek's efficient reasoning model with strong coding and math capabilities.
deepseek/deepseek-v3.2 DeepSeek's fast and efficient V4 model optimized for speed with a 1 million token context window.
deepseek/deepseek-v4-flash DeepSeek's flagship V4 reasoning model with strong coding and math capabilities and a 1 million token context window.
deepseek/deepseek-v4-pro Moonshot AI
Moonshot AI's Kimi models — long-context reasoning.
Moonshot AI's advanced reasoning model with strong performance on complex tasks.
moonshotai/kimi-k2.5 Z-AI
Z-AI's GLM models — fast, multilingual general-purpose intelligence.
Z-AI's fast and efficient general-purpose model with strong multilingual capabilities.
zai/glm-5-turbo MiniMax
MiniMax's general-purpose models with large context support.
MiniMax's latest general-purpose model with strong reasoning and large context support.
minimax/minimax-m2.7 Qwen
Alibaba's Qwen models — high-performance reasoning with up to 1M-token context.
Qwen's flagship model with state-of-the-art reasoning and a 1 million token context window.
alibaba/qwen3.7-max Qwen's high-performance model with 1 million token context window.
alibaba/qwen3.5-plus-02-15 Qwen's fast and efficient model optimized for speed with 1M context.
alibaba/qwen3.5-flash-02-23