DeepCoder-14B-Preview is a code reasoning LLM fine-tuned from DeepSeek-R1-Distilled-Qwen-14B using distributed reinforcement learning (RL) to scale up to long context lengths.

Excels in complex reasoning tasks like math and coding. Uses reinforcement learning to match larger models' performance while remaining efficient.

Phi-4-mini-instruct is a lightweight open model built upon synthetic data and filtered publicly available websites - with a focus on high-quality, reasoning dense data. The model belongs to the Phi-4 model family and supports 128K token context length.

Efficient Mixture-of-Experts model with 37B active parameters per token, excelling in coding, math, and reasoning tasks while maintaining 128K token context length.

microsoft/Phi-3-mini-4k-instruct

BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation

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

OpenCoder is an open and reproducible code LLM family which includes 1.5B and 8B base and chat models, supporting both English and Chinese languages.

Anthropic's best combination of performance and speed for efficient, high-throughput tasks.

Given a prompt and/or an input image, the model will generate a new image.

Handles a range of natural language processing tasks, offering strong text generation, translation, summarization, and code capabilities across multiple domains.

Qwen2.5 is the latest series of Qwen large language models. For Qwen2.5, we release a number of base language models and instruction-tuned language models ranging from 0.5 to 72 billion parameters. Qwen2.5 brings the following improvements upon Qwen2:

Fast, Lightweight Model

A specialized, cost-efficient AI model focused on STEM reasoning, delivering strong performance in mathematics and coding at faster speeds.

Excels at multi-step reasoning, advanced math, coding, and science by spending more time thinking and leveraging chain-of-thought rationales for complex tasks.

Llama 3.1 8B

Llama 3.1 is an auto-regressive language model that uses an optimized transformer architecture.

Delivers strong text and vision capabilities with low cost and latency—ideal for real-time, large context, or multitask AI applications.

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

Text embeddings

GPT-4o is our most advanced multimodal model that’s faster and cheaper than GPT-4 Turbo with stronger vision capabilities. The model has 128K context and an October 2023 knowledge cutoff.

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

Cost-efficient, fast, and reliable option for use cases such as translation, summarization, and sentiment analysis.

State-of-the-art Mistral model trained specifically for code tasks.

Powerful model for on-device use cases.

Version-capable small model.

State-of-the-art Mistral model trained specifically for code tasks.

This is broken, just for testing

General Text Embeddings (GTE) model. Towards General Text Embeddings with Multi-stage Contrastive Learning. The GTE models are trained by Alibaba DAMO Academy. They are mainly based on the BERT framework

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

The BAAI/bge-m3 is an advanced multilingual embedding model supporting over 100 languages, designed for high-performance semantic similarity and retrieval tasks. By generating dense vector representations that capture semantic meaning across diverse linguistic contexts, the model enables sophisticated cross-lingual text understanding and comparison.

A fast 8B model

A 7B transformer model, fast-deployed and easily customisable.

A 7B sparse Mixture-of-Experts (SMoE). Uses 12.9B active parameters out of 45B total.

meta-llama/Meta-Llama-3-8B-Instruct-Turbo

Qwen2-0.5B-Instruct is a compact 500 million parameter language model from Alibaba's Qwen series, designed for efficient instruction-following and conversational tasks with a lightweight architecture optimized for resource-constrained environments.

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 405B is the latest flagship model in the Hermes series of LLMs by Nous Research, and the first full parameter finetune since the release of Llama-3.1 405B.

OpenAI’s text embeddings measure the relatedness of text strings.

Specializes in code generation, reasoning, and fixing with 128K token context, open-source licensing, and local deployment capabilities.

Mixtral 8x22B is currently the most performant open model. A 22B sparse Mixture-of-Experts (SMoE). Uses only 39B active parameters out of 141B.

Qwen2-7B-Instruct is an advanced 7 billion parameter large language model from Alibaba's Qwen series, optimized for instructional and conversational tasks with strong multilingual capabilities and improved performance across reasoning, mathematics, and coding benchmarks. The model builds on the Qwen series' reputation for high-quality open-source AI models, offering enhanced instruction-following abilities and efficient text generation across multiple domains.

<needs_description>

Excels at complex multilingual reasoning, code generation, and precise instruction following, with native fluency in five languages and advanced function calling.