End-to-end platform for developing your AI applications

No matter your starting point, let's build aligntag

Book a Demo Contact Sales

Explore AI Models Directory

All categories and models you can try out and seamlessly integrate in your projects

Featured

openai/whisper-large-v3-turbo

Whisper is a state-of-the-art model for automatic speech recognition (ASR) and speech translation, proposed in the paper "Robust Speech Recognition via Large-Scale Weak Supervision" by Alec Radford et al. from OpenAI. Trained on >5M hours of labeled data, Whisper demonstrates a strong ability to generalise to many datasets and domains in a zero-shot setting. Whisper large-v3-turbo is a finetuned version of a pruned Whisper large-v3. In other words, it's the exact same model, except that the number of decoding layers have reduced from 32 to 4. As a result, the model is way faster, at the expense of a minor quality degradation.

Featured

Ministral-3-14B-Instruct-2512

Ministral 3 14B Instruct (2512) is the largest model in the Ministral 3 family, delivering frontier-level capabilities with performance comparable to the larger Mistral Small 3.2 24B model. It is a powerful yet efficient multimodal language model, offering strong text understanding alongside vision capabilities. This model is the instruction-tuned, post-trained Instruct variant in FP8, optimized to follow user instructions accurately. It is well suited for chat-based interactions and a wide range of instruction-driven use cases. Designed with edge deployment in mind, the Ministral 3 family supports execution across diverse hardware environments. Ministral 3 14B can be deployed locally, fitting within 24GB of VRAM in FP8—and even less when further quantized—making it a practical choice for on-device and resource-constrained deployments.

14B

Featured

Llama-3.3-70B-Instruct

Meta Llama 3.3 is a multilingual large language model (LLM) with 70 billion parameters, designed for text-in/text-out generation and instruction-tuned performance. The Llama 3.3 70B Instruct model is optimized for multilingual conversational use cases and demonstrates strong performance, outperforming many open-source and proprietary chat models on widely used industry benchmarks.

70B

Llama-4-Maverick-17B-128E-Instruct

Llama 4 Maverick is part of the Llama 4 family, a new generation of natively multimodal AI models designed to support rich text and multimodal experiences. Built on a mixture-of-experts architecture, Llama 4 models deliver industry-leading performance in both text and image understanding. The Llama 4 release marks the beginning of a new era for the Llama ecosystem. This series introduces two efficient models: Llama 4 Scout, a 17-billion-parameter model with 16 experts, and Llama 4 Maverick, a 17-billion-parameter model featuring 128 experts, offering enhanced capacity and scalability.

Mistralai Devstral-2-123B-Instruct-2512

Devstral 2 123B Instruct (2512) is an agentic large language model designed specifically for software engineering tasks. It excels at using tools to explore large codebases, edit multiple files, and power advanced software engineering agents. The model achieves outstanding performance on SWE-bench, demonstrating its strong real-world coding capabilities. This FP8 Instruct model has been fine-tuned to follow instructions, making it well suited for chat-based interactions, agentic workflows, and instruction-driven software engineering use cases. For enterprises that require specialized capabilities—such as extended context lengths or domain-specific knowledge—we encourage organizations to contact us to discuss customized solutions.

123B

Featured

Qwen3-VL-235B-A22B-Thinking

Qwen3-VL is the most powerful vision-language model in the Qwen series to date. This generation delivers comprehensive improvements across the board, including stronger text understanding and generation, deeper visual perception and reasoning, extended context length, enhanced spatial and video dynamics comprehension, and more robust agent interaction capabilities. Qwen3-VL is available in both Dense and Mixture-of-Experts (MoE) architectures, scaling seamlessly from edge to cloud deployments. It also offers Instruct and reasoning-enhanced Thinking editions, enabling flexible, on-demand deployment for a wide range of use cases.

235B

Qwen3-Embedding-0.6B

Qwen3-Embedding-0.6B is part of the Qwen3 Embedding model series, the latest proprietary embedding and reranking models in the Qwen family, purpose-built for text embedding and ranking tasks. Built on the dense foundational models of the Qwen3 series, it delivers strong multilingual performance, long-text understanding, and reasoning capabilities in a compact and efficient model size. The Qwen3 Embedding series provides a comprehensive suite of embedding and reranking models ranging from 0.6B to 8B parameters, supporting a wide range of applications such as text and code retrieval, text classification, clustering, and bitext mining. Despite its lightweight footprint, Qwen3-Embedding-0.6B inherits the advanced multilingual and cross-lingual capabilities of its larger counterparts, supporting over 100 languages and multiple programming languages. Designed for flexibility and practical deployment, the Qwen3 Embedding models support customizable vector dimensions and user-defined instructions, enabling developers to optimize performance for specific tasks, languages, or scenarios. This makes Qwen3-Embedding-0.6B an ideal choice for efficiency-focused applications that require high-quality embeddings without sacrificing scalability or versatility.

0.6B

Z-Image-Turbo

Z-Image-Turbo is a high-efficiency image generation foundation model with 6 billion parameters, designed to deliver strong performance without relying on extremely large model sizes. Serving as the core model in the Z-Image ecosystem, it demonstrates that top-tier image generation quality can be achieved through systematic architectural and training optimizations rather than sheer scale. Z-Image-Turbo excels in photorealistic image synthesis and bilingual text rendering, achieving results that are comparable to leading commercial image generation models. Its efficient design makes it well suited for practical deployment scenarios where performance, cost, and scalability must be carefully balanced. As a central component of the Z-Image project, Z-Image-Turbo represents a compelling step forward in efficient visual foundation models, combining strong generation quality with a compact and deployable model footprint.

Featured

stable-diffusion-xl-nsfw

Ready to Accelerate AI in Your Organization?

Contact our sales team to discuss your enterprise needs and deployment options.

Get Started