Browse available models routable through Multi-Router. Compare capabilities, pricing, and specs.
The fastest model with near-frontier intelligence. Best for low-latency, high-volume workloads.
The most intelligent model for building agents and coding. Exceptional performance in reasoning, complex analysis, and multi-step tasks.
The best combination of speed and intelligence. Ideal for high-throughput tasks requiring strong reasoning.
Best price-performance model for large scale processing, low-latency, high-volume tasks with thinking.
Fastest and most budget-friendly multimodal model. Ideal for classification, extraction, and low-latency workloads.
State-of-the-art thinking model for complex problems in code, math, STEM, and long-context analysis.
Frontier-class reasoning and multimodal understanding at a fraction of the cost of larger models.
Most cost-efficient multimodal model. Optimized for high-frequency lightweight tasks like translation and data extraction.
Advanced intelligence with better thinking, improved token efficiency, and factual consistency. Optimized for software engineering and agentic workflows.
General-purpose model with tool use and 197K context. Free tier.
Compact MoE model for efficient text generation with 256K context. Free tier.
MoE model (120B total, 12B active) optimized for text tasks with 262K context. Free tier.
Compact vision-language model with 128K context. Free tier.
Compact reasoning model with 128K context. Free tier.
Excels at instruction following and tool calling with broad knowledge across domains. 1M token context.
Strong coding, reasoning, and agentic capabilities across domains with built-in reasoning tokens.
Near-frontier intelligence for cost-sensitive, low-latency, high-volume workloads.
Fastest and cheapest GPT-5 variant. Optimized for summarization and classification tasks.
Most capable frontier model. Exceptional at complex professional work with configurable reasoning depth.
Premium tier using extended compute for harder problems. Best for research-grade reasoning tasks.
Sets a new standard for math, science, coding, and visual reasoning tasks with highest reasoning capability.
Extended compute version of o3 for harder problems. Best for research-grade reasoning tasks.
Fast, cost-efficient reasoning model optimized for coding and visual tasks.
MoE coding model (480B total, 35B active) with tool use and 262K context. Free tier.
MoE instruction-tuned model with tool use and 262K context. Free tier.
MoE model (196B total, 11B active) with strong reasoning and 256K context. Free tier.
Reasoning model with 131K context from Zhipu AI. Free tier.
Still available but consider migrating to current models for improved performance.