Alibaba Cloud Provider

Wan3.0 is an All-in-One reference-based video generation model that supports Text-to-Video, Image-to-Video (first frame/first-last frame), and Reference-based Video Generation. It can generate videos up to 30 seconds long at 30fps.

Wan3.0-Video-Prime is the high-speed version of Wan3.0’s video generation model. It matches the capabilities of the standard Wan3.0-Video model, supports comprehensive reference inputs across four modalities, generates videos up to 30 seconds long, and delivers immersive audiovisual experiences with significantly faster end-to-end performance.

Qwen-Audio-3.0-TTS-Plus is Alibaba's higher-quality text-to-speech model, generating spoken audio from text via the DashScope Speech Synthesizer API.

Qwen-Audio-3.1-ASR-Flash is a large-scale model supporting efficient short speech recognition, designed for high-quality, multilingual, and cultural content transcription scenarios. The model supports multilingual and multi-regional Chinese dialect recognition, and is optimized for the rhythm, cadence, and classical Chinese expression of ancient Chinese poetry. It features context enhancement, punctuation prediction, and text normalization capabilities. This version also supports controllable ASR/AST output for multiple dialects. It includes native transcription and polishing capabilities, as well as industrial-grade multi-person dialogue role-based transcription capabilities.

The Qwen3-VL-Embedding model series are the latest additions to the Qwen family, built upon the recently open-sourced and powerful Qwen3-VL foundation model. Specifically designed for multimodal information retrieval and cross-modal understanding, this suite accepts diverse inputs including text, images, screenshots, and videos, as well as inputs containing a mixture of these modalities.

Qwen3 Rerank is a new proprietary model of the Qwen model family. Specifically designed for reranking tasks, built on the Qwen3 foundation model. Leveraging Qwen3’s robust multilingual text understanding capabilities, the series achieves state-of-the-art performance across multiple benchmarks for text embedding and reranking tasks.

The Qwen3-VL-Rerank model series are the latest additions to the Qwen family, built upon the recently open-sourced and powerful Qwen3-VL foundation model. Specifically designed for multimodal information retrieval and cross-modal understanding, this suite accepts diverse inputs including text, images, screenshots, and videos, as well as inputs containing a mixture of these modalities.

Happy Horse 1.0 is described as an open-source state-of-the-art AI video generator with native joint audio-video generation — meaning the Happy Horse AI video model produces video frames and the corresponding audio track (dialogue, ambient sound, Foley) together in a single forward pass, rather than generating silent video and dubbing it afterward. According to community-compiled architecture notes, the model is built around a 15-billion-parameter unified self-attention Transformer that processes text, image, video, and audio tokens within a single token sequence. It is reportedly built without dedicated cross-attention branches and without a separate audio module. Combined with DMD-2 distillation, the distilled variant is reported to generate 1080p video in roughly 38 seconds on an NVIDIA H100, using only 8 denoising steps without classifier-free guidance.

Qwen3.7-Max is the flagship model in Alibaba's Qwen3.7 series. It supports text input and output and is designed for agent-centric workloads, with particular strengths in coding, office and productivity tasks, and long-horizon autonomous execution. The model offers notable gains in coding and agentic performance over prior Qwen generations and supports explicit prompt caching for efficient repeated context use.

Qwen3.6-Plus is Alibaba’s next-generation Qwen large language model released on April 2, 2026. Compared with version 3.5, Qwen 3.6 has made significant overall performance improvements and has exhibited remarkably strong agent-oriented programming capabilities.

Qwen3.5 native vision-language Flash model series, based on a hybrid architecture design, integrates linear attention mechanisms with sparse Mixture-of-Experts (MoE) models, achieving higher inference efficiency. The model's performance has made a leap forward compared to the 3 series in both pure text and multimodal capabilities; it offers fast response times, balancing inference speed and performance.

Qwen3.7 Flash is a vision-language reasoning model from Alibaba. It is suited for multimodal agents, visual coding, search, and computer interaction, with strengths in object recognition, spatial understanding, and real-world visual perception.

Qwen3.7-Plus is a multimodal generation model in the Qwen3.7 family that supports text, image, and video inputs with text output. It is designed for agent-oriented workloads and strong performance in coding, office productivity, and long-horizon autonomous execution.

Qwen3-ASR-Flash is a highly accurate, intelligent, and robust multilingual speech recognition model based on a large language model. Leveraging a powerful foundational model, massive amounts of text and multimodal data, and tens of millions of hours of audio data, Qwen3-ASR-Flash achieves high-precision speech recognition. It can automatically determine the language and accurately recognize speech in 11 languages, ensuring precise transcription even in complex audio environments.

2.4-trillion-parameter MoE flagship delivering a comprehensive leap in coding and professional work. Autonomously codes and delivers complete projects spanning 10+ days. Handles hundreds of specialized tasks across legal, financial, design, and other professional domains, producing production-grade results end-to-end in a single conversation. Native visual understanding runs through the full cycle of planning, execution, and verification, enabling deep semantic analysis of ultra-long documents and extended video content. In long-horizon tasks, plans autonomously, iterates through closed feedback loops, and continuously evolves.

Qwen3.6 Flash is a fast, efficient language model from Alibaba's Qwen 3.6 series. It supports text, image, and video input with a 1M token context window. Tiered pricing kicks in above 256K tokens. Prompt caching is supported, with both explicit cache read and cache creation pricing.

Qwen3.5 Native Visual Language Series Plus model, based on a hybrid architecture design, integrates linear attention mechanisms with sparse mixture-of-experts models, achieving higher reasoning efficiency. In multiple task evaluations, the 3.5 series has demonstrated outstanding performance comparable to current top-tier cutting-edge models, with model performance showing leaps and bounds of progress compared to the 3 series in both pure text and multimodal aspects.

Qwen3.8-Flash is the latest multimodal model from the Qwen family, combining powerful reasoning and generation with remarkable speed. It natively supports a million-token context window, allowing it to process lengthy documents, entire codebases, and complex conversations in a single pass. It shines in coding assistance, agentic workflows, and visual understanding — whether it's fixing code autonomously, operating desktop applications, or analyzing charts and long videos. Fully compatible with both OpenAI and Anthropic API protocols, it integrates seamlessly with popular developer tools like Claude Code and Codex, making it easy to build high-concurrency applications and intelligent workflows. With strong performance and highly competitive inference costs, Qwen3.8-Flash is an ideal choice for developers and businesses seeking the best of both worlds in AI applications.

Qwen3.8-2.4T-A95B is the open-source version of Tongyi Qianwen's latest flagship series, released in August 2026. It adopts a sparse MoE architecture with a total of 2.4 trillion parameters, activating approximately 95 billion per step, and supports 1 million token contexts with a hybrid attention mechanism. Core benchmarks include GPQA Diamond 92.6, PaperBench 93.0, OSWorld 86.1, and BabyVision 82.0. It ranks 4th globally on CodeArena.

Qwen3.8-Max-0902 is a snapshot release of Qwen3.8-Max with stronger coding performance for complex engineering tasks and long-horizon autonomous development. It also improves agentic tool use and visual understanding for chart reasoning, document parsing, and multimodal perception, while retaining a 1M-token context window, reasoning mode, and a full tool ecosystem.

The Qwen3.8 series 27B native vision-language dense model delivers significant improvements over the 3.6-27B version, with a focused enhancement in programming and office-scenario capabilities across both text and visual modalities. It can more reliably complete complex tasks end-to-end and produce trustworthy outputs.

Designed for audio and video understanding and content analysis, it supports text, image, audio, and video inputs with text output.

The Qwen3 series VL models effectively integrates thinking and non-thinking modes, achieving world-leading performance in visual agent capabilities on public benchmark datasets such as OS World. This version features comprehensive upgrades in areas like visual coding, spatial perception, and multimodal reasoning, significantly enhancing visual perception and recognition abilities, and supporting the understanding of ultra-long videos.

GLM-5.3-Prime is the high-speed variant of Z.ai's GLM-5.3, inheriting its full capabilities while delivering 1.5–2× higher output throughput through inference acceleration. It supports text input and output with a 1M-token context window and up to 128K output tokens, and is optimized for coding and agentic workloads, including long-horizon multi-turn agent orchestration, real-time conversation, and streaming code generation.

In addition to DeepSeek, TencentCloud and AlibabaCloud now also offer the official 0731 version, while other providers are still using the old 0424 version. . DeepSeek V4 Pro is DeepSeek’s flagship open-weight Mixture-of-Experts model, released in preview on April 24, 2026. It features 1.6 trillion total parameters, with 49 billion activated per token, and supports a one-million-token context window. Designed for advanced reasoning, coding, STEM tasks, and autonomous agents, it delivers state-of-the-art performance among open models and rivals leading proprietary systems. The model supports both thinking and non-thinking modes and is available through DeepSeek’s applications and APIs.

DeepSeek V4.1 Flash is a 552B-parameter MoE model built on a new causal encoder-decoder architecture, designed for higher capability, faster reasoning, higher throughput, and lower serving cost. It natively supports multimodal visual understanding and delivers flagship-level intelligence with significantly reduced KV cache requirements, making it well suited for high-throughput and cost-sensitive agentic workloads.

GLM-5.2 supports a larger context window and output length, with up to 1M context and 128K output tokens. It adds streaming output for tool calls via tool_stream=true, enabling real-time access to tool call arguments. Like the GLM-4.7 series, it supports deep thinking with thinking={ type: "enabled" }, where thinking is enforced once enabled, and introduces reasoning_effort to control reasoning intensity under chain-of-thought mode. It also delivers stronger coding performance and more advanced reasoning capabilities.

GLM-5.3 is Zhipu AI’s next-generation flagship text model, built on GLM-5.2 and enhanced through large-scale, long-horizon task post-training. It delivers an approximately 50% improvement in coding experience, autonomously handling complex project development, debugging, validation, and delivery. It also supports a 1-million-token context window, up to 128K-token outputs, tool calling, and MCP, with strong performance in cybersecurity tasks such as code review and vulnerability detection.

Qwen-Image-2.0-Pro is an image generation model launched by Qwen AI.

Rich content: Supports input of up to 4.5k tokens and dense information layout with images-within-images, enabling complex layouts like newspapers, storyboards, menus, and exam papers to be generated in a single pass. Authentic detail: Supports precise rendering of text as small as 10px, and vividly reproduces fine details such as micro-expressions, pores, and individual strands of hair—approaching the quality of real photography. Deep knowledge: Supports native rendering of 12 languages and 20+ fonts, realistic simulation of mainstream interfaces such as web pages, games, and live streams, fully incorporating external knowledge. Qwen-Image-3.0-Pro isn't just pursuing "good looks"—it's pursuing "usefulness", making image generation a truly deployable productivity tool.

Accurate Prompt Understanding: Supports up to 4.5k token inputs, accurately interpreting complex text-and-image prompts and generating dense layouts in one pass. Reliable Text Rendering: Delivers crisp 10px text across 12 languages and 20+ fonts, making infographics and interfaces ready to use. Efficient Batch Production: Scales everyday tasks like posters, web pages, and UI screens with better cost efficiency, keeping creative output sustainable. Qwen-Image-3.0 Standard is built not just for visual quality, but for smooth, reliable daily creation—turning image generation into sustainable content productivity.

Qwen-Image-2.0 is an image generation model launched by Qwen AI.