Модели

Ultra-fast, cheapest GPT model for high-volume tasks and real-time applications.

Free, lightweight model for basic tasks and experimentation.

Free model for testing and prototyping with basic capabilities.

Compact version of GPT-4.1 optimized for speed and cost efficiency.

Compact and fast model for lightweight tasks. Great balance of speed and capability.

Fast and affordable model for lightweight tasks. Great for high-throughput applications.

Fast and efficient model for high-throughput tasks at low cost.

Small but mighty model from Microsoft, punching above its weight in reasoning tasks.

Open-weight model with excellent performance-to-cost ratio. Great for self-hosting.

Efficient model for straightforward tasks with good multilingual support.

Latest GPT model with improved instruction following and coding capabilities.

Lightweight open model with vision capabilities. Great for self-hosting.

Best balance of intelligence, speed, and cost. Excellent for most enterprise tasks.

Llama 3.3 served on Groq hardware for ultra-low latency inference.

Next-gen open model with MoE architecture. 128 experts, strong multilingual performance.

Most advanced multimodal model. Great for complex tasks requiring vision, analysis, and creative output.

Specialized coding model with state-of-the-art code generation and completion.

Google's most capable model with 1M context and strong reasoning capabilities.

Cost-effective general-purpose model with strong coding and analysis capabilities.

Mixture of Experts model with excellent performance across diverse tasks.

Mistral's flagship model for complex reasoning with strong multilingual support.

Powerful open model from Alibaba with strong multilingual and coding capabilities.

Reasoning model optimized for STEM tasks. Uses chain-of-thought for complex problems.

Scalable model optimized for RAG workflows and enterprise use cases.

Reasoning model with chain-of-thought capabilities. Excellent for math and logic.

Most capable model for complex reasoning and analysis tasks. Extended thinking enabled.

NVIDIA's large model optimized for helpfulness and alignment.

xAI's latest model with strong reasoning and conversational abilities.

Enterprise-grade RAG model with excellent retrieval-augmented generation.

Advanced reasoning model that thinks before answering. Best for science and math.

Search-augmented model with real-time web access. Great for research tasks.