ONE API. FRONTIER MODELS.
One unified, OpenAI-compatible API for fast, cost-efficient access to frontier models.
Built for developers. Open to possibility.
Keep requests moving when a provider is unavailable.
Understand performance and troubleshoot individual requests.
Zero data retention on eligible models.
EXPLORE THE POSSIBILITIES
From reasoning to moving images, discover what you can build with Musube. Choose a modality to explore the model families behind it.
ONE API. EVERY DIRECTION.
Reason, write, and build with a new generation of models.
01THE MODEL COLLECTION
Discover the models shaping China’s AI ecosystem, alongside a broader catalog. Choose by what you want to build.
From complex questions to considered answers. Put reasoning to work.
deepseek-v4-pro-260425High-end DeepSeek reasoning for planning and complex analysis.
deepseek-v4-flash-260425Low-latency general model for high-throughput prompting and drafting.
deepseek-v4-flash-0731Pinned DeepSeek Flash snapshot for reproducible, version-locked output.
A versatile foundation for conversations, agents, and everyday intelligence.
qwen3.8-maxThe newest Qwen Max generation, the strongest Alibaba tier for reasoning and agents.
qwen3.7-maxQwen's most capable tier for hard reasoning and agentic tasks.
qwen3.7-max-previewEarly access to the next Qwen Max, latest capabilities first.
qwen3.7-max-2026-06-08Pinned dated snapshot of Qwen 3.7 Max for reproducible output.
qwen3.7-plusBalanced Qwen tier for quality prompting at lower cost.
qwen3.6-max-previewStrong long-context reasoning for scripts and storyboards.
qwen3.6-27bCompact open-weight Qwen for cost-sensitive, high-volume workloads.
qwen3.6-flashLow-latency Qwen for high-volume drafting and classification.
Connect the dots across long documents and demanding reasoning tasks.
glm-5.2Strong long-context reasoning for structured scripts and storyboards.
glm-5.2-fast-previewPreview of the low-latency GLM tier for high-throughput prompting.
Go from understanding a codebase to building what comes next.
kimi-k2.7-codeLong-context Kimi tuned for code generation and repo-scale reasoning.
Turn a prompt or a still into motion, with a model built for video.
doubao-seedance-2-5-260628The latest Seedance generation, with the strongest motion coherence and prompt adherence in the family.
doubao-seedance-2-0-260128Flagship text→video and image→video with synced audio. The default route for new video projects.
doubao-seedance-2-0-fast-260128Lower-latency Seedance 2.0 for high-volume short-form where speed beats polish.
doubao-seedance-2-0-mini-260615The cheapest Seedance tier for drafts and high-throughput iteration.
Explore cinematic movement, visual storytelling, and new perspectives.
kling-v3-omni-t2vKling v3 Omni text-to-video for cinematic, high-motion shots.
kling-v3-omni-i2vKling v3 Omni image-to-video with strong subject consistency.
kling-v3-t2vStandard Kling v3 text-to-video, in std and pro modes for 720p or 1080p output.
kling-v3-i2vStandard Kling v3 image-to-video for animating stills with cinematic camera moves.
02ROOM TO EXPERIMENT
Buy credits. Choose your model. Pay for what you use. Put your budget behind the experiments worth making.
Let’s talk about your workloadUse credits across the models on Musube. Explore a new model without starting a new provider relationship.
Pricing depends on the model and modality. We’ll help you understand the rates and billing units for your workload.
03FROM IDEA TO INFERENCE
A focused path from finding the right model to making it part of your product.
Access is currently by invitation. Share your use case and the models you want to work with.
We’ll help you get access, understand model pricing, and add credits to your account.
Create your API key, choose a model, and connect Musube to your application.
api.musube.ai04A LITTLE MORE CONTEXT
Working on something specific?
We’d like to hear about it.
Musube brings affordable, reliable inference to one unified, OpenAI-compatible API. Access frontier models across reasoning, images, video, and audio without managing a separate integration for every provider.
Access is currently by invitation. Contact sales@musube.ai with a short description of your project and the models you’re interested in. We’ll help you with availability, pricing, and onboarding.
The model collection above shows our catalog across text, video, image, and audio. Expand a model family to see its model IDs. Talk to us to confirm availability and the right configuration for your workload.
Musube uses credits for usage-based billing. Rates and billing units depend on the model and modality. We’ll confirm the applicable pricing with you during onboarding, so you can plan around your workload.
Musube provides an OpenAI-compatible API for supported workflows. The request format and features depend on the model and modality, especially for image, video, and audio generation. We’ll help you confirm the integration for your application.