Back

AI provider / Alibaba Cloud

Alibaba Cloud

A provider frequently referenced for broad model ranges, practical enterprise-cloud framing and model lines discussed in open-weight contexts.

Alibaba Cloud now reads less like a single Qwen story and more like a full first-party AI catalog inside Model Studio. The current lineup spans Qwen3.8 flagship and flash text models, coding agents, OCR, image generation and editing, video generation, speech, realtime translation, multimodal Omni variants, retrieval layers such as embeddings and rerankers, decision/classification models and, with HappyOyster, explorable world models.

This page intentionally tracks Alibaba-owned lines such as Qwen, Wan, HappyHorse, HappyOyster and CosyVoice, rather than third-party models (DeepSeek, GLM, Kimi) that also sit in the platform catalog. That makes the provider easier to read as a producer in its own right: one where cloud infrastructure, agent tooling and model breadth are designed to reinforce each other.

Model catalog

Every release in this provider’s model line.

Each row is a distinct model release — its date, its type and the traits it is known for. Open a source to dig into any of them.

ModelReleasedTypeKey traitsSource
Qwen-Image-2.1-TurboOct 2026Visual
  • Cost-effective
  • Compact architecture
  • Unified gen + edit
Qwen-Image-2.1-ProOct 2026Visual
  • 7B unified gen + edit
  • Native RGBA
  • Up to 10 reference images
Qwen Decision ModelSep 2026Preview
  • Classification + scoring
  • Parallel decisions
  • Confidence scores
Qwen3.8-Omni-Flash-RealtimeSep 2026Multimodal
  • Realtime audio + video
  • Remote MCP tools
  • WebSocket + WebRTC
Qwen3.8-Omni-FlashSep 2026Multimodal
  • Up to 1M tokens
  • OpenAI + Anthropic compatible
  • Spatial audio
HappyOyster 1.0Sep 2026World model
  • Adventure, Directing, Acting
  • Explorable open worlds
  • Realtime interaction
Qwen3.8-LiveTranslate-Flash-RealtimeSep 2026Audio
  • 60 input languages
  • 29 output languages
  • ~2.3s latency
Qwen-Flash-CharacterAug 2026Role-play
  • Character consistency
  • Context-aware dialogue
  • Empathetic tone
Qwen-MT-Image 2.0Aug 2026Translation
  • 55 languages
  • Terminology control
  • Subject detection
Qwen3.8-FlashAug 2026Cost optimized
  • 1M-token context
  • Multimodal
  • Rivals Anthropic, DeepSeek
Wan 3.0 VideoAug 2026WAN
  • All-in-One generation
  • Prime fast variant
  • Audio included
Qwen3.8-27BAug 2026Visual
  • Native VL dense
  • 27B
  • Coding + office tasks
Qwen3.8-2.4T-A95BAug 2026Open release
  • 2.4T params open
  • ~95B active
  • Thinking mode
Qwen3.7-Text-EmbeddingAug 2026Embeddings
  • 256-2560 dims
  • Multilingual
  • +20% MTEB retrieval
Qwen-Audio-3.0-RealtimeAug 2026Audio
  • Plus + Flash tiers
  • Duplex conversation
  • Speech reasoning
Qwen-Image-3.0Aug 2026Visual
  • Pro + base tiers
  • Dense layouts
  • 10px text rendering
Qwen3.8-MaxAug 2026Flagship
  • 2.4T MoE params
  • 1M-token context
  • Long-horizon coding
Qwen-Audio-3.0-ASRJul 2026Audio
  • 7 dialect groups
  • 30 languages
  • Streaming + file
Qwen3.7-FlashJul 2026Cost optimized
  • Native VL
  • Upgraded multimodal agent
  • Low latency
Qwen-Audio-3.0-TTSJul 2026Audio
  • Plus + Flash tiers
  • Chinese dialects
  • Minority languages
HappyHorse 1.1Jun 2026WAN
  • T2V + I2V + R2V
  • 3-15s at 1080P
  • Audio generation
Qwen3.7-MaxJun 2026Flagship
  • Top Qwen 3.7
  • Agent productivity
  • Autonomous execution
Qwen3.7-PlusJun 2026Flagship
  • Multimodal agent
  • GUI + mobile
  • Automation
Qwen3.5-LiveTranslate-Flash-RealtimeMay 2026Audio
  • 60 input languages
  • 29 output languages
  • Live translation
Qwen-Image-2.0-ProApr 2026Visual
  • Highest quality
  • Text fidelity
  • Texture detail
HappyHorse 1.0Apr 2026Visual
  • I2V + T2V
  • Fluid motion
  • Reference consistency
Qwen3.6-FlashApr 2026Cost optimized
  • Native VL
  • Math + coding
  • Spatial intelligence
Qwen3.6-Max PreviewApr 2026Flagship
  • Vibe coding
  • Efficient execution
  • Frontend dev
Wan 2.7 Video SuiteApr 2026WAN
  • R2V + edit
  • Multi-reference
  • Style transfer
Qwen3.6-PlusApr 2026Flagship
  • Native VL
  • Agent coding
  • OCR + vision
Qwen3.5-Omni-FlashMar 2026Multimodal
  • 10h audio
  • 400s video
  • Function calling
Qwen-Image-2.0Mar 2026Visual
  • Generate + edit
  • 1000-token text
  • Fast pipeline
Qwen3-ASR-FlashMar 2026Audio
  • ASR
  • Auto language detect
  • Multilingual
CosyVoice V3 PlusFeb 2026Audio
  • Human-like TTS
  • Natural prosody
  • Next-gen voice
Qwen3-RerankJan 2026Embeddings
  • RAG
  • 100+ languages
  • Long documents
Qwen-Image-Edit-MaxJan 2026Visual
  • Image editing
  • Character consistency
  • LoRA support
Qwen3-TTS-VD-RealtimeJan 2026Audio
  • Realtime TTS
  • 11 languages
  • Designed voices
Qwen-Image-MaxDec 2025Visual
  • Image generation
  • Realistic textures
  • Better text
Qwen3-VL-PlusDec 2025Visual
  • Visual coding
  • Long video
  • Spatial perception
Wan 2.6-T2IDec 2025WAN
  • Text to image
  • Photoreal textures
  • Text rendering
Qwen3-Omni-FlashDec 2025Multimodal
  • Text-image-audio-video
  • 119 languages
  • Streaming voice
Qwen3-TTS-FlashNov 2025Audio
  • Low latency
  • 17 voices
  • Multilingual
Qwen-VL-OCRNov 2025Visual
  • OCR
  • Document parsing
  • Image text
Qwen3-MaxSep 2025Flagship
  • Agent upgrade
  • Tool invocation
  • Complex workflows
Qwen3-Coder-PlusSep 2025Coding
  • Coding agent
  • Tool use
  • Autonomous workflows
Qwen3-LiveTranslate-Flash-RealtimeSep 2025Audio
  • Realtime translation
  • 19 languages
  • Audio + video
Qwen-EmbeddingAug 2025Embeddings
  • 64-2048 dims
  • Multilingual
  • Retrieval
Qwen-FlashAug 2025Cost optimized
  • 1M context
  • Thinking toggle
  • Low-cost tier
Qwen-MT-PlusJul 2025Translation
  • 92 languages
  • Terminology control
  • Context-aware
Qwen3Apr 2025Hybrid reasoning
  • Thinking modes
  • Multiple sizes
  • Open weight
Qwen2.5-MaxJan 2025Flagship
  • Large-scale MoE
  • Proprietary
  • Benchmark push
QwQ-32B-PreviewNov 2024Reasoning
  • Chain-of-thought
  • 32B
  • Apache 2.0
Qwen 2.5Sep 2024Model release
  • 100+ open models
  • Coding
  • Math
Qwen2Jun 2024Model release
  • 27 languages
  • 128K context
  • MoE variants
Qwen 1.5Feb 2024Open release
  • 0.5B-72B
  • 6 open sizes
  • Edge to datacenter
Tongyi Qianwen 2.0Oct 2023Version update
  • Hundreds of B params
  • Industry models
  • Stronger reasoning
Tongyi Qianwen (Qwen)Apr 2023Debut
  • Multi-turn chat
  • Multimodal
  • Cloud-native