Models
Browse 24 canonical LLM models across all providers
Anthropic's latest and most advanced model with state-of-the-art reasoning, coding, and analysis capabilities.
Anthropic's balanced model offering strong performance with efficient cost for everyday tasks.
Anthropic's previous flagship model with excellent reasoning and coding capabilities.
OpenAI's most capable model with advanced reasoning, multimodal understanding, and tool use.
OpenAI's fifth-generation model with strong general intelligence and tool use.
Google's most advanced multimodal model with native audio, video, and code understanding.
Google's powerful multimodal model with long context and strong reasoning.
xAI's flagship model with real-time knowledge, strong reasoning, and humor.
xAI's speed-optimized model for low-latency inference with strong capabilities.
Moonshot AI's advanced model with strong multilingual and long-context capabilities.
Meta's largest open-weight model with 400B parameters and state-of-the-art open performance.
Meta's efficient open-weight model optimized for deployment with strong capabilities.
Meta's efficient 70B parameter model with strong instruction following and coding.
Mistral AI's flagship model with strong multilingual, reasoning, and coding capabilities.
Mistral AI's balanced model offering strong performance at moderate cost.
NVIDIA's large-scale model optimized for enterprise AI workloads and reasoning.
DeepSeek's most capable model with advanced reasoning and coding at competitive pricing.
DeepSeek's advanced reasoning model with strong coding capabilities at low cost.
Alibaba's specialized coding model with strong code generation and understanding.
Mistral's specialized coding model built for software development workflows.
OpenAI's most capable model for complex real-world work including coding, research, and document creation.
Meta Superintelligence Labs' first model with advanced reasoning, multimodal understanding, and agentic capabilities.
Alibaba's efficient MoE model with 35B total / 3B active parameters. Frontier-level agentic coding performance.
Sber's flagship MoE model with 702B total / 36B active parameters for multilingual workloads and reasoning.