Model Library
Llama 4 Maverick 17B (128E) Instruct
meta-llama/Llama-4-Maverick-17B-128E-Instruct-FP8
High-capacity multimodal language model built on a mixture-of-experts (MoE) architecture with 128 experts and 17 billion active parameters. Supports multilingual text and image input.
Deepseek V3 03-24
deepseek-ai/DeepSeek-V3-0324
An iteration of the DeepSeek V3 model with notable improvements in reasoning capabilities across various benchmarks, including MMLU-Pro, GPQA, and AIME.
Llama 4 Scout 17B (16E) Instruct
meta-llama/Llama-4-Scout-17B-16E-Instruct
A mixture-of-experts (MoE) language model activating 17 billion parameters out of a total of 109B. Designed for assistant-style interaction and visual reasoning.
DeepSeek R1 (Basic)
deepseek-ai/DeepSeek-R1
DeepSeek R1 is a powerful reasoning model designed for complex problem-solving and advanced coding tasks.
Qwen2-VL-7B-Instruct
An advanced vision-language model. It is purpose-built for tasks that involve both visual and textual understanding, including image captioning, visual question answering, and content generation.
Serverless
Qwen2-VL-2B-Instruct
A powerful multimodal model that excels in visual understanding tasks, including image and video comprehension.
Qwen2.5-VL-7B-Instruct
A 7B-parameter multimodal model, designed for complex vision-language tasks. Ideal for document understanding, video summarization, and interactive applications requiring both visual perception and language reasoning.
Serverless
Llama 4 Scout 17B (16E) Instruct
A mixture-of-experts (MoE) language model activating 17 billion parameters out of a total of 109B. Designed for assistant-style interaction and visual reasoning.
Serverless
