High quality edge model with only 3B parameters. Test this model through our UI, and then download the weights and run it on the edge on edge devices.

Powerful edge model with extremely high performance/price ratio.

A 12B model with image understanding capabilities in addition to text.

Multilingual open source model.

Decoder-only transformer for chat based purposes.

Enterprise-grade 22B parameter small model with the lastest version v2 released September 2024.

Cutting-edge language model for coding.

Sparse Mixture-of-Experts (MoE) model with a total of 45 billion parameters. Best model overall regarding cost/performance trade-offs.

Reasoning model for high-complexity tasks with the lastest version v2 released July 2024.

Sparse Mixture-of-Experts (SMoE) model that uses only 39B active parameters out of 141B, offering unparalleled cost efficiency for its size.

Smaller text embeddings with a dimension of

Larger text embeddings

Affordable and intelligent small model for fast, lightweight tasks

High-intelligence flagship model for complex, multistep tasks.

More efficient than o1-preview with similar reasoning capabilities.

Language model trained with reinforcement learning to perform complex reasoning.

The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models with 8B

The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models with 70B

Qwen 2.5 Coder is the latest series of code-specific Qwen large language models.

Qwen2.5 are a series of decoder-only language models developed by Qwen team, Alibaba

The 72B variant of the latest iteration of Qwen-VL model from Alibaba, representing nearly a year of innovation.

A a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token from Deepseek.

Llama 3.3 70B Instruct is the December update of Llama 3.1 70B. The model improves upon Llama 3.1 70B (released July 2024) with advances in tool calling, multilingual text support, math and coding. The model achieves industry leading results in reasoning, math and instruction following and provides similar performance as 3.1 405B but with significant speed and cost improvements.

Qwen QwQ model focuses on advancing AI reasoning, and showcases the power of open models to match closed frontier model performance.QwQ-32B-Preview is an experimental release, comparable to o1 and surpassing GPT-4o and Claude 3.5 Sonnet on analytical and reasoning abilities across GPQA, AIME, MATH-500 and LiveCodeBench benchmarks.

The Meta Llama 3.1 collection of multilingual large language models (LLMs) is a collection of pretrained and instruction tuned generative models with 405B

DeepSeek-R1 is a state-of-the-art large language model optimized with reinforcement learning and cold-start data for exceptional reasoning, math, and code performance.

Llama 3.2 3B Instruct is a lightweight, multilingual model from Meta. The model is designed for efficiency and offers substantial latency and cost improvements compared to larger models.

8 billion parameter instruct-tuned decoder-only handles complex language tasks with high accuracy and efficiency.

Qwen2.5-Coder is the latest series of Code-Specific Qwen large language models. It has excellent capability for its size.

70 billion parameter instruct-tuned decoder-only handles complex language tasks with high accuracy and efficiency.

More powerful than Llama 3.1 70B

Llama 3.1 Nemotron 70B Instruct is a large language model fine tuned by NVIDIA to improve the helpfulness of LLM generated responses to user queries. It scores better than the original Llama 3.1 70B on benchmarks and in the LLM Arena.

Qwen QwQ model focuses on advancing AI reasoning, and showcases the power of open models to match closed frontier model performance. QwQ 32B Preview is an experimental release, comparable to o1 and surpassing GPT-4o and Claude 3.5 Sonnet on analytical and reasoning abilities across GPQA, AIME, MATH-500 and LiveCodeBench benchmarks.

A a strong Mixture-of-Experts (MoE) language model with 671B total parameters with 37B activated for each token from Deepseek.

405 billion parameter instruct-tuned decoder-only handles complex language tasks with high accuracy and efficiency.

DeepSeek-R1 is a state-of-the-art large language model optimized with reinforcement learning and cold-start data for exceptional reasoning, math, and code performance.

Gemini API generates state-of-the-art embeddings for words, phrases, and sentences

High volume and lower intelligence tasks

Our fastest multimodal model with great performance for diverse, repetitive tasks and a 1 million context window. Now generally available for production

Gemini 1.5 Pro is a mid-size multimodal model that is optimized for a wide-range of reasoning tasks.

Llama 3.2 1B is an extremely small, fast, and cheap model from Meta that is very capable for its size.

Llama 3.2 3B instruct is a lightweight, multilingual model from Meta. The model is designed for efficiency and offers substantial latency and cost improvements compared to larger models.

Llama 3.1 8B is a flexible model from Meta that is very capable for its size.

A powerful multimodal model capable of processing both text and image inputs that supports multilingual, multi-turn conversations, tool use, and JSON mode.

Gemma 2 9B Instruct is a lightweight open model from Google, built from the same research and technology used to create the Gemini models.

The Meta Llama 3.3 multilingual large language model (LLM) is a pretrained and instruction tuned generative model in 70B (text in/text out).

Llama 3.3 70B Instruct is the December update of Llama 3.1 70B. The model improves upon Llama 3.1 70B (released July 2024) with advances in tool calling, multilingual text support, math and coding. The model achieves industry leading results in reasoning, math and instruction following and provides similar performance as 3.1 405B but with significant speed and cost improvements.

A powerful multimodal model capable of processing both text and image inputs that supports multilingual, multi-turn conversations, tool use, and JSON mode.

Hermes 3 is the latest version of the flagship Hermes series of LLMs by Nous Research

Hermes 3 is the latest version of the flagship Hermes series of LLMs by Nous Research

Hermes 3 is the latest version of the flagship Hermes series of LLMs by Nous Research