Available models | Oxen.ai

Home Repositories Models Docs Blog Pricing Login Sign up

Search

Model Library

Choose the right model, get to the perfect prompt.

New models added every week.

;

Copyright © 2025 Oxen Labs, Inc., All Rights Reserved

Careers Privacy Policy Terms and Conditions

Model Library

Choose the right model, get to the perfect prompt.

108 models. New models added every week.

Search Models

Model Types

Fine-tuning

Fine-tunable

22

Modalities

Image to Image

11

Image to Text

25

Image to Video

4

Text to Embeddings

3

Text to Image

10

Text to Text

78

Text to Video

7

Video to Text

1

Video to Video

1

Developers

Openai

22

Qwen

19

Google

14

Meta

10

Mistral

9

Anthropic

8

Black_forest_labs

5

Deepseek

5

Perplexity

5

Alibaba

4

ByteDance

2

Moonshotai

2

Tongyi-MAI

1

Kling

1

CMU

1

Qwen-Image-Edit-2511

Delivers high-fidelity, controllable image editing with dual semantic and appearance modes, precise on-image text, multi-image composition, and robust identity preservation.

Gemini 3 Flash

Fast multimodal model with configurable reasoning, strong agentic workflows, long context, and tool use for interactive chat, coding, and complex tasks.

text-to-textimage-to-textvideo-to-text

WAN 2.6 - Image to Video

Animates images into 15s, 1080p videos with preserved identity, native audio, lip-sync, and multi-shot sequences guided by reference videos.

text-to-videoimage-to-video

GPT Image 1.5

Diffusion model for high‑fidelity image generation and editing, with strong prompt adherence, preserved composition and lighting, and adjustable quality controls.

Seedream 4.5

High-fidelity text-to-image and image-to-image generation with multi-reference control (up to 10 images), 4K support, and batch output.(128 characters)

Z-Image-Turbo

Fast photorealistic text-to-image model with accurate English and Chinese on-image text, ideal for interactive design, marketing visuals, and UI/UX workflows.

FLUX.2 [dev]

Black Forest Labs

Generates photorealistic images with precise multi-reference editing, excels at legible text and infographics, and supports rapid LoRA fine-tuning workflows.

text-to-imageimage-to-image

FLUX.2 [pro]

Black Forest Labs

Delivers photorealistic, high-resolution images with advanced multi-reference editing, precise pose and color control, and reliable prompt and text adherence for professionals.

text-to-imageimage-to-image

FLUX.2 [flex]

Black Forest Labs

Delivers high-quality image generation and editing with advanced text rendering, multi-image reference for style consistency, and precise, JSON-based prompt control.

text-to-imageimage-to-image

Claude Opus 4.5

Excels at long-horizon reasoning, advanced coding, dynamic effort control, robust multimodal tasks, and detailed computer interface inspection for complex workflows.

text-to-textimage-to-text

Nano Banana Pro

Delivers high-fidelity images with advanced text rendering, consistent character identities, and precise prompt following for professional visual design and branding.

Segment Anything 3 - Image

Zero-shot image segmentation with text/visual prompts; exhaustive instance detection and presence head reduce false positives.

Segment Anything 3 - Video

Detects, segments, and tracks objects across video frames using text, exemplars, points, or masks, with memory for occlusions and real-time streaming.

Gemini 3 Pro Preview

Excels at deep multimodal reasoning, complex coding, and advanced tool use across text, images, audio, and video with a vast 1M token context.

text-to-textimage-to-text

GPT 5.1

Automatically routes prompts to fast or deep reasoning modes, with adaptive effort, enhanced tone and style controls, and improved coding and math.

text-to-textimage-to-text

Kimi K2 Thinking

Enables autonomous, step-by-step reasoning and orchestration of 200–300 tool calls for complex research, coding, and web tasks across long contexts.

Veo 3.1

Generates high-fidelity videos with native synced audio, offering strong narrative control, scene consistency, image-to-video animation, and multi-shot support.

text-to-videoimage-to-video

Gemini 2.5 Flash Preview

Delivers rapid, cost-effective multimodal responses with flexible reasoning, supports text, image, and native audio output, and adapts style or tone via prompts.

text-to-textimage-to-text

Gemini 2.5 Flash Lite Preview

Optimized for rapid, high-volume multimodal tasks with a 1M-token context window, delivering strong reasoning and cost efficiency for enterprise workflows.

text-to-textimage-to-text

Qwen-Image-Edit-Plus

Delivers high-fidelity, controllable image editing with dual semantic and appearance modes, precise on-image text, multi-image composition, and robust identity preservation.

Kling v2.5 - Image to Video

Transforms single images into smooth, cinematic videos with natural motion, realistic camera work like dolly zooms, and preserved style. (108 chars)

text-to-videoimage-to-video

nano-banana

Generates photorealistic images with precise prompt and text rendering, mask-free editing, and layout-aware outpainting, ideal for creative and multilingual content.

Seedream 4.0

Delivers ultra-fast, high-resolution image generation, precise natural-language editing, and consistent multi-image output—ideal for creative, batch, or professional workflows.

Claude Sonnet 4.5

Anthropic's most advanced AI model, excelling in coding, agent-based tasks, and computer usage. It delivers high performance in reasoning, math, and domain-specific knowledge across fields like finance, law, and STEM.

text-to-textimage-to-text

Qwen-Image-Edit

Enables precise bilingual text and semantic edits with strong consistency, advanced multi-image editing, and native pose/control support for creative compositions.

Qwen3 VL 8B - Instruct

versatile multimodal large language model capable of understanding and generating both text and images. Built on the Qwen3 architecture, it provides strong general reasoning, detailed image interpretation, and instruction-following performance in a compact 8B parameter size.

text-to-textimage-to-text

Qwen3 VL 2B - Instruct

Lightweight multimodal model for visual Q&A, multilingual OCR, document and UI understanding, and agentic screen interpretation in constrained environments.

text-to-textimage-to-text

Qwen3 VL 4B - Instruct

Multimodal LLM for text and images, excelling in visual QA, document/UI understanding, spatial reasoning, image captioning, and multimodal coding.

text-to-textimage-to-text

FLUX.1 [dev]

Black Forest Labs

Open-weight text-to-image model with advanced prompt adherence, anatomically accurate details, and powerful tools for inpainting, outpainting, and structural edits.

GPT 5 Mini

Optimized for cost and speed, handles long contexts, supports text and image input, and excels at structured outputs and tool integration for precise tasks.

GPT 5 Nano

Multimodal model optimized for ultra-fast, cost-efficient summarization and classification, supporting both text and image inputs with real-time streaming output.

GPT 5

Handles complex reasoning, code generation, and multimodal inputs with improved accuracy, long context retention, and robust multilingual and personalization features.

OpenAI/GPT-OSS-20B

Delivers strong reasoning and chain-of-thought, agentic features, and multilingual support, optimized for local deployment and efficient use on modest hardware.

OpenAI GPT OSS 120B

Built with a Mixture-of-Experts design, delivers efficient, transparent reasoning, tool use, and agentic capabilities, even with 128K token context windows.

Claude Opus 4.1

Excels at complex coding, autonomous research, and agent workflows, with advanced reasoning and a 200,000-token context for deep analysis and synthesis.

Qwen-Image

Excels at multilingual text rendering and precise image editing, supporting ControlNet guidance and diverse styles from photorealistic to anime and minimalist.

Wan2.2 A14B - Text to Video

Delivers high-fidelity text-to-video synthesis at 480p/720p using dual expert models for scene layout and fine motion detail, ideal for creative production.

Qwen 3 Coder 480B

Designed for complex coding and agentic workflows, it excels at understanding large codebases and supports 262K-token context for advanced software engineering.

Qwen3 Coder 480B (A35B) Instruct

Enables advanced agentic coding, large-scale code analysis, and tool integration with up to 256K native context for repository-level tasks and automation.

Kimi K2 Instruct

Handles complex reasoning, coding, and tool use with 128K token context, enabling autonomous problem-solving and workflow automation.

Gemini 2.5 Flash

Fast, cost-efficient multimodal reasoning model with million-token context for high-volume applications requiring speed and versatility.

text-to-textimage-to-text

FLUX.1-Kontext [dev]

Black Forest Labs

Delivers precise, iterative image editing and generation with consistent character, style, and text changes—using multimodal input for seamless scene transformations.

Claude Opus 4

Excels at deep reasoning, complex coding, and autonomous agent workflows with sustained performance, extended thinking, tool use, and memory across tasks.

Claude Sonnet 4

Balances intelligence with efficiency for coding, research, and automation tasks; excels in reasoning, content generation, and nuanced instruction following.

text-to-textimage-to-text

Veo 3.0

Generates realistic text- and image-conditioned videos with native synchronized audio, including dialogue, ambient sound, and effects.

text-to-videoimage-to-video

Gemini 2.5 Pro

Excels at building interactive web apps, advanced code editing and agentic workflows, with native multimodality and strong video-to-code capabilities.

text-to-textimage-to-text

Qwen/Qwen3-4B

Dual reasoning modes enable rapid or step-by-step responses, with robust support for over 100 languages and long-context processing up to 262,144 tokens.

Qwen 3 235B-A22B

MoE model with 22B active parameters featuring dual thinking modes for complex reasoning and efficient conversation across 100+ languages.

Qwen 3 30B-A3B

MoE architecture with 3.3B active parameters, balancing efficiency with strong reasoning, multilingual capabilities, and specialized thinking mode.

Qwen/Qwen3-0.6B

Efficient conversational AI for resource-limited devices with multilingual support, document summarization, translation, code generation, and simple information retrieval.

o4 mini

Optimized for fast, affordable reasoning with strong coding and visual skills, large 200k-token context, and efficient handling of complex tasks.

text-to-textimage-to-text

o3

Excels at advanced reasoning, coding, math, and visual tasks with simulated reasoning, tool use, web browsing, and image understanding integration.

text-to-textimage-to-text

GPT 4.1 mini

Powerful mid-sized model with GPT-4o-level performance at lower cost and latency, featuring a 1 million token context window for complex tasks.

text-to-textimage-to-text

GPT 4.1

Excels in coding and instruction following with million-token context window, enabling superior performance on complex, multi-step tasks.

text-to-textimage-to-text

GPT 4.1 nano

OpenAI's fastest, cost-effective model with full 1 million token context, optimized for classification, autocompletion, and real-time AI agent tasks.

text-to-textimage-to-text

Llama 4 Scout

Multimodal model specializing in multilingual text and image analysis, multi-document summarization, codebase reasoning, and highly personalized task automation.

Llama 4 Maverick

Multimodal model with 17B active parameters, excelling at text, image, code, and multilingual tasks. Supports 1M-token context for advanced enterprise use.

text-to-textimage-to-text

Qwen2.5-VL 32B Instruct

Handles advanced visual recognition, complex analysis of images and videos, structured data extraction, agentic tool use, and robust multilingual reasoning.

Deepseek V3

Delivers advanced reasoning, code generation, and mathematical skills, processes long inputs efficiently, and accelerates results with innovative Mixture-of-Experts design.

Mistral Small 3.1

A lightweight, versatile 24B multimodal model handling text and images with extensive multilingual support and 128k token context window.

Gemma 3 27B

Gemma 3 has a large, 128K context window, multilingual support in over 140 languages, and is available in more sizes than previous versions. Gemma 3 models are well-suited for a variety of text generation and image understanding tasks, including question answering, summarization, and reasoning.

Perplexity Sonar Deep Research

Performs exhaustive, multi-step research by autonomously searching and synthesizing hundreds of sources into detailed, expert-level reports across domains.

Perplexity Sonar Reasoning Pro

Premium reasoning model for complex, multi-step analysis. Delivers detailed explanations, real-time web search, and double citations for thorough answers.

Llama 3.2 1B Instruct

Efficient, multilingual instruction-tuned model designed for privacy-focused, on-device dialogue, summarization, and agentic retrieval across mobile and edge platforms.

Wan2.1 14B - Text to Video

Generates high-fidelity, temporally consistent videos from text or images, with readable English and Chinese text, sound effects, and customizable aspect ratios.

Claude 3.7 Sonnet

A hybrid reasoning model with standard and extended thinking modes, delivering twice the speed and exceptional performance in coding and problem-solving tasks.

text-to-textimage-to-text

Perplexity Sonar Reasoning

Fast reasoning model with real-time web search, chain-of-thought capabilities, and citation support. Excels at complex queries with quick, accurate responses.

Perplexity Sonar

Optimized for search-augmented tasks, delivering fast, accurate answers with real-time web data and detailed citations. Excels in research and fact-checking.

Perplexity Sonar Pro

Excels at complex, multi-step queries with real-time web search, detailed answers, extensive citations, and customizable information retrieval.

Deepseek R1

Employs a massive Mixture-of-Experts architecture and Multi-Layer Attention to deliver advanced, polished reasoning and problem-solving across math, code, and more.

Deepseek R1

An open-source reasoning model using Mixture-of-Experts architecture, delivering powerful math and code capabilities comparable to OpenAI's o1.

Deepseek V3

Powers advanced reasoning, code generation, and multilingual tasks with efficient MoE architecture and enhanced multi-token prediction for faster, optimized results.

Llama 3.3 70B Instruct

Optimized for dialogue with strong reasoning, multilingual support, and efficient performance approaching larger models.

QwQ 32B Preview

Specialized in advanced reasoning and problem-solving, excelling in mathematics and programming with a 32B parameter transformer architecture.

Qwen 2.5 Coder 32B Instruct

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

Llama 3.2 3B Instruct

Optimized for multilingual dialogue, agentic tasks, and efficient on-device use, supporting eight languages and a 128K context for privacy-focused applications.

Claude 3.5 Sonnet

Powerful AI with exceptional coding abilities, twice the speed of previous versions, and advanced reasoning for complex software development tasks.

Claude 3.5 Haiku

Anthropic's fastest model offering advanced coding, tool use, and reasoning capabilities with rapid response times for real-time applications and personalized experiences.

Ministral 8B

Efficient edge model with native function calling and interleaved sliding-window attention for fast, memory-efficient processing in resource-constrained environments.

Qwen2.5 1.5B Instruct

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:

Qwen2.5 72B Instruct

Instruction-tuned LLM excelling in long-context processing (131K tokens), multilingual support (29+ languages), and structured data handling.

o1 preview

Reasoning-focused LLM for complex science, math, and coding tasks, generating detailed thought processes before responses.

o1 mini

Optimized for coding and math with Chain-of-Thought reasoning, offering fast, cost-efficient responses for complex problem-solving.

Pixtral 12B

Multimodal model handling text and images at native resolution with 128K context window, excelling in visual reasoning tasks like document analysis and image captioning.

Llama 3.1 70B Instruct

Multilingual LLM excelling in question answering, reasoning, code generation, and synthetic data generation.

Llama 3.1 8B Instruct

Optimized for multilingual dialogue and instruction following, supporting eight major languages for chat, text generation, and translation tasks.

Llama 3.1 405B Instruct

Optimized for multilingual dialogue with 128k context, instruction-tuned via SFT/RLHF, and enhanced with synthetic data for safety and performance.

Mistral Nemo

Handles long-form content with 128k token context, excels in multilingual tasks, coding, and function calling via natural language.

GPT 4o mini

Cost-efficient, fast model with 128K context window, supporting text/vision inputs and improved multilingual performance.

text-to-textimage-to-text

Text Embedding 004

Generates vector representations capturing semantic meaning/context for tasks like semantic search, text classification, and clustering. Multilingual support with versatile applications.

text-to-embeddings

GPT 4o

Multimodal LLM for real-time text, audio, and visual processing with multilingual support, emotional audio responses, and image generation.

text-to-textimage-to-text

Mixtral 8x22B

Efficient Sparse MoE architecture with 39B active parameters, excels in multilingual tasks, math, coding, and handles 64K token contexts.

Mistral Large 2

Powerful LLM with 123B parameters, excelling in multilingual tasks, coding, and reasoning, optimized for single-node inference and long-context applications.

Text Embedding 3 - Large

Generates high-quality embeddings for complex text analysis and multilingual applications with 8,191 token context.

text-to-embeddings

Text Embedding 3 - Small

Generates compact, efficient embeddings for NLP tasks with multilingual support, balancing performance and low latency.

text-to-embeddings

Mixtral 8x7B

Efficient Mixture of Experts (8 experts) with 13B active parameters, optimized for multilingual tasks and cost-performance balance.

DALL-E 3

Translates nuanced text prompts into detailed, accurate images with automatic prompt rewriting, multiple aspect ratios, and ChatGPT integration for creative workflows[1][2][6].

Mistral 7B

Balanced performance in natural language and code tasks, efficiently handling longer sequences with innovative attention mechanisms.

Open Pose

OpenPose is a real-time computer vision model that detects and maps body, hand, and facial keypoints for multiple people simultaneously using convolutional neural networks.

Wan2.1 1.3B - Text to Video

Generates 480P videos from text prompts on consumer GPUs, with multilingual support, image-to-video, aspect ratio control, and audio integration features.

Qwen/Qwen3-1.7B

Efficiently generates multilingual text and code, with dual modes for rapid chat or detailed reasoning; ideal for lightweight AI, agents, and education.

Gemini 2.0 Flash

Multimodal LLM for agentic applications, handling real-time data integration and multi-step tasks with enhanced reasoning via Thinking Mode, integrating Google tools and third-party functions.

text-to-textimage-to-text

Ministral 3B

Optimized for edge computing with function-calling capabilities, excelling in knowledge retrieval and commonsense reasoning with 128k token context.

o1

Specializes in complex reasoning through chain-of-thought processing, excelling in STEM tasks like coding, math, and scientific analysis.

Deepseek R1 Distill Llama 70B

Delivers strong mathematical and coding abilities, matching the performance of larger models while using efficient distillation and multilingual support.

GPT 4.5

Excels in natural conversation and creative tasks with improved emotional intelligence and multilingual support, prioritizing intuitive interactions over structured reasoning.

Gemini 2.0 Flash Lite

Cost-efficient, budget-friendly multimodal LLM for real-time tasks with 1M token input context and enhanced performance.

text-to-textimage-to-text

o3 mini

Optimized for STEM reasoning and problem-solving, excelling in complex tasks like advanced math and coding with improved cost efficiency.