Vendor · Language models · Image models · Video models · Audio models
Alibaba has 67 models in the directory. The strongest is Qwen 3.8 Max 0902: Rank 22 to 32 of 95 in the overall index.
In the directory: 59 language models, 3 image models, 3 video models and 2 audio models. 54 of them with open weights that can run locally.
Every model there ever was. Models with a profile lead to every benchmark, price and source.
| Qwen3.8-Omni-Flash | Sep 17, 2026 | Active | n/a | $0.11 / $0.38 | n/a | 1M | 2 |
| Qwen 3.8 Max 0902 | Sep 2, 2026 | API only | 50Rank 22 to 32 of 95 | $1.65 / $4.95 | 41 | 1M | 68 |
| Qwen 3.8 Flash | Aug 26, 2026 | API only | n/a | $0.11 / $0.38 | n/a | 1M | 62 |
| Qwen 3.8 Flash Next | Aug 26, 2026 | Active | n/a | n/a | 55 | 262K | 61 |
| Qwen 3.8 27B | Aug 14, 2026 | Active | 11Rank 44 to 57 of 95 | $0.04 / $2.50 | 40 | 262K | 70 |
| Qwen 3.8 2.4T A95B | Aug 12, 2026 | Active | n/a | $2 / $6 | 41 | 262K | 42 |
| Qwen 3.7 Flash | Jul 15, 2026 | API only | n/a | $0.028 / $0.11 | n/a | 1M | 46 |
| Qwen 3.7 Plus | June 2026 | API only | 13Rank 44 to 56 of 95 | $0.28 / $1.10 | 58 | 1M | 24 |
| Qwen 3.7 Max | May 2026 | API only | 17Rank 43 to 52 of 95 | $1.65 / $4.95 | 203 | 1M | 23 |
| Qwen 3.6 27B | Apr 21, 2026 | Active | n/a | $0.30 / $3.20 | 58 | 262K | 51 |
| Qwen 3.6 35B A3B | Apr 17, 2026 | Active | n/a | $0.10 / $0.95 | 141 | 262K | 51 |
| Qwen 3.6 Max-Previewno profile | April 2026 | No status data | n/a | n/a | n/a | n/a | n/a |
| Qwen3.5 0.8B | Mar 2, 2026 | Active | n/a | n/a | n/a | 262K | 72 |
| Qwen3.5 2B | Mar 2, 2026 | Active | n/a | n/a | n/a | 262K | 71 |
| Qwen3.5 4B | Mar 2, 2026 | Active | n/a | n/a | 18 | 262K | 76 |
| Qwen 3.5 9B | Feb 27, 2026 | Active | n/a | $0.10 / $0.15 | 57 | 262K | 56 |
| Qwen 3.5 122B A10B | Feb 24, 2026 | Active | n/a | $0.26 / $2.08 | 132 | 262K | 51 |
| Qwen 3.5 27B | Feb 24, 2026 | Active | n/a | $0.27 / $2.16 | 77 | 262K | 45 |
| Qwen 3.5 35B A3B | Feb 24, 2026 | Active | n/a | $0.15 / $1 | 153 | 262K | 43 |
| Qwen 3.5 397B A17B | Feb 15, 2026 | Active | n/a | $0.55 / $3.50 | 59 | 262K | 48 |
| Qwen 3 Coder Next | Feb 4, 2026 | Active | n/a | $0.12 / $0.80 | 84 | 262K | 41 |
| Qwen 3 VL 32B | Oct 19, 2025 | Active | n/a | n/a | 68 | 262K | 45 |
| Qwen 3 VL 8B | Oct 11, 2025 | Active | n/a | $0.25 / $0.75 | 120 | 262K | 45 |
| Qwen 3 VL 8B Thinking | Oct 11, 2025 | Active | n/a | n/a | 113 | 262K | 30 |
| Qwen 3 VL 30B A3B | Sep 30, 2025 | Active | n/a | $0.20 / $0.70 | 23 | 262K | 36 |
| Qwen 3 VL 30B A3B Thinking | Sep 30, 2025 | Active | n/a | $0.29 / $1 | 111 | 262K | 46 |
| Qwen3 VL 235B A22B Instruct | Sep 23, 2025 | Active | n/a | n/a | 46 | 131K | 28 |
| Qwen 3 VL 235B A22B | Sep 22, 2025 | Active | n/a | $0.21 / $1.90 | 49 | 262K | below 1 |
| Qwen 3 VL 235B A22B Thinking | Sep 22, 2025 | Active | n/a | n/a | 56 | 262K | 53 |
| Qwen3 Omni 30B A3B Thinking | Sep 22, 2025 | Active | n/a | n/a | 102 | 66K | 60 |
| Qwen 3 Next 80B A3B Instruct | Sep 9, 2025 | Active | n/a | $0.09 / $1.10 | 134 | 262K | 34 |
| Qwen 3 Next 80B A3B Thinking | Sep 9, 2025 | Active | n/a | $0.15 / $1.20 | 194 | 262K | 31 |
| Qwen 3 Coder 30B A3B | Jul 31, 2025 | Active | n/a | $0.07 / $0.28 | 89 | 262K | 24 |
| Qwen3-Coder 30B-A3B Instruct | Jul 31, 2025 | Active | n/a | n/a | n/a | 262K | 20 |
| Qwen 3 30B A3B Thinking 2507 | Jul 29, 2025 | Active | n/a | n/a | 150 | 262K | 32 |
| Qwen 3 30B A3B Instruct 2507 | Jul 28, 2025 | Active | n/a | $0.09 / $0.30 | 192 | 262K | 49 |
| Qwen 3 235B A22B Thinking 2507 | Jul 25, 2025 | Active | n/a | $0.45 / $3.50 | 131 | 262K | 28 |
| Qwen 3 Coder 480B A35B | Jul 23, 2025 | Active | n/a | $0.30 / $1 | 59 | 262K | 43 |
| Qwen3-Coder 480B-A35B Instruct | Jul 22, 2025 | Active | n/a | n/a | n/a | 262K | 33 |
| Qwen 3 235B A22B Instruct 2507 | Jul 21, 2025 | Active | n/a | $0.15 / $0.75 | 48 | 262K | 41 |
| Qwen 3 14B | Apr 27, 2025 | Active | n/a | $0.10 / $0.22 | 71 | 33K | 48 |
| Qwen 3 235B A22B | Apr 27, 2025 | Active | n/a | n/a | 60 | 33K | 43 |
| Qwen 3 30B A3B | Apr 27, 2025 | Active | n/a | $0.12 / $0.50 | 102 | 33K | 49 |
| Qwen 3 32B | Apr 27, 2025 | Active | n/a | $0.14 / $0.40 | 52 | 33K | 44 |
| Qwen 3 8B | Apr 27, 2025 | Active | n/a | n/a | 39 | 33K | 50 |
| QwQ 32B | Mar 5, 2025 | Active | n/a | n/a | n/a | 131K | 45 |
| Qwen 2.5 VL 72B | Jan 27, 2025 | Active | n/a | $0.80 / $1 | n/a | 128K | 22 |
| Qwen2.5-VL 72B Instruct | Jan 26, 2025 | Active | n/a | n/a | n/a | 131K | 28 |
| Qwen2.5-Coder-0.5B | Nov 12, 2024 | Active | n/a | n/a | n/a | 33K | 39 |
| Qwen2.5-Coder-32B-Instruct | Nov 12, 2024 | Active | n/a | n/a | n/a | 131K | 33 |
| Qwen 2.5 Coder 32B | Nov 6, 2024 | Active | n/a | $0.66 / $1 | n/a | 131K | 19 |
| Qwen 2.5 7B | Sep 16, 2024 | Active | n/a | $0.10 / $0.20 | n/a | 131K | 49 |
| Qwen 2.5 0.5Bno profile | September 2024 | No status data | n/a | n/a | n/a | n/a | n/a |
| Qwen 2.5 14Bno profile | September 2024 | No status data | n/a | n/a | n/a | n/a | n/a |
| Qwen 2.5 32Bno profile | September 2024 | No status data | n/a | n/a | n/a | n/a | n/a |
| Qwen 2.5 72B | September 2024 | Active | n/a | $0.36 / $0.40 | n/a | 131K | 17 |
| Qwen2.5-32B-Instruct | September 2024 | Active | n/a | n/a | n/a | 131K | 49 |
| Qwen2.5-72B-Instruct | September 2024 | Active | n/a | n/a | n/a | 131K | 30 |
| Qwen2-72B-Instruct | June 2024 | Active | n/a | n/a | n/a | 131K | 32 |
| Qwen-Image-2.1 | Sep 20, 2026 | Active | n/a | 3 |
| Qwen Image 2.0 Pro | April 2026 | Active | $0.075 per image | 34 |
| Wan 2.6 Text to Image | January 2026 | Active | $0.03 per image | 13 |
| Wan 2.7 | April 2026 | Active | $9 per minute | 30 |
| Wan 2.6 | December 2025 | Active | $9 per minute | 24 |
| Wan 2.2 A14B | July 2025 | Active | $4.80 per minute | 27 |
| Qwen3.8-Omni-Flash-Realtime | Sep 21, 2026 | Active | n/a | below 1 |
| Qwen3.8-LiveTranslate | Sep 18, 2026 | Active | n/a | 1 |
Audio model · Sep 21, 2026
Qwen3.8-Omni-Flash-Realtime connects live text, audio, and video input to text and audio output. Its realtime documentation lists WebSocket, WebRTC, and AOQ transports; function calling and MCP form the documented action path.
Image model · Sep 20, 2026
Qwen-Image-2.1 combines text-to-image generation and editing in one local open-weight model with 7B parameters. Up to ten reference images, RGBA output with transparent layers, and local editing controls based on circles, painted annotations, and separate masks define its concrete focus.
Audio model · Sep 18, 2026
Qwen3.8-LiveTranslate targets real-time interpretation rather than general text chat. It translates audio and images while producing text and audio, supports 60 languages, and can speak 29 of them. Speaker separation, synchronized bilingual output, and long-context disambiguation matter when a conversation has multiple voices and the attribution matters as much as the words.
Language model · Sep 17, 2026
Qwen3.8-Omni-Flash is a proprietary API model for audio and video analysis with text output. It accepts text, images, audio, and video in a shared context of up to 1,000,000 tokens. The documentation names long recordings, meeting summaries, speaker alignment, and video understanding.
Language model · Sep 2, 2026
New in Qwen 3.8 Max 0902 is the combination of a 2.4-trillion-parameter MoE with Alibaba's focus on long coding and office tasks. The documentation also highlights visual understanding, chart reasoning, and document parsing. The snapshot accepts text, images, and video.
Language model · Aug 26, 2026
What distinguishes Qwen 3.8 Flash is the combination of multimodal input and a 1,000,000-token window in an API. Alibaba documents text, images, and video as inputs and text as output. In Thinking mode, the API exposes up to 262,144 tokens of chain of thought.
Language model · Aug 26, 2026
Qwen 3.8 Flash Next is an open experimental architecture checkpoint, not simply a smaller Qwen3.8 model. Qwen replaces Gated Attention with Qwen Sparse Attention and combines it with Gated DeltaNet. The model has 125 billion parameters, activates 6 billion, and adds 51 billion n-gram embeddings. It accepts text, images, and video and uses Thinking mode by default.
Language model · Aug 14, 2026
Qwen 3.8 27B is a dense open-weight vision-language model for text, images, and video. Its model card combines a vision encoder with a hybrid Gated DeltaNet and Gated Attention architecture.
Language model · Aug 12, 2026
Qwen 3.8 2.4T A95B is the open-weight checkpoint behind the Qwen3.8-Max serving release. Its repository contains FP8-quantized weights for a hybrid Mixture-of-Experts model with 2.4 trillion total parameters and 95 billion active parameters. The architecture combines Gated DeltaNet and Gated Attention.
Language model · Jul 15, 2026
Qwen 3.7 Flash concentrates its 1,000,000-token configuration on long image and video analysis. The proprietary endpoint accepts text, images, and video and processes videos up to two hours or 2 GB. It allows 256 images per request, compared with 2,048 for Qwen 3.8 and Qwen 3.7 Plus. The limit of 64 videos is unchanged.
Language model · June 2026
What distinguishes Qwen 3.7 Plus is Alibaba's hybrid-agent design. It reads screens, operates graphical interfaces, writes code from visual references, and combines GUI with CLI steps. Text, images, and video are inputs; the context window is 1,000,000 tokens.
Language model · May 2026
Qwen 3.7 Max combines a 1,000,000-token window with Alibaba's focus on coding, office automation, MCP integrations, and long autonomous runs. Its integrated thinking mode can be controlled through the API. The parameter count remains undisclosed, and the weights are not downloadable.
51 more current models are in the model list above.
Derived from the overall index, the API price and the measured speed of the current models.
The strongest choice
Qwen 3.8 Max 0902
The vendor's highest overall index
Value for money
Qwen 3.7 Plus
Costs 19% of Qwen 3.8 Max 0902, overall index 13 instead of 50
The fastest
Qwen 3.7 Max
203 tokens/s, a typical model manages 98
The 20 best language models, with the models from Alibaba highlighted. “All” and the filter show every model in the index.
20 of 95 models
Provider
Weights
Price (blended per 1M tokens)
Released
Anthropic
OpenAI
Meta
Moonshot AI
DeepSeek
xAI
Z.ai
Xiaomi
Alibaba
StepFun
Mistral AI
MiniMax
Thinking Machines Lab
Inception
| Vendor | Models | Model types | Best language model | Top 10 | Open weights | New in 2026 |
|---|---|---|---|---|---|---|
| Anthropic | 33 | 33 language models | Claude Opus 5.5 (Rank 1 to 3) | 4 | 0 | 13 |
| OpenAI | 63 | 46 language models, 4 image models, 2 video models, 9 audio models, 2 specialist models | GPT-6 Astra (Rank 4 to 8) | 2 | 4 | 27 |
| 66 | 43 language models, 4 image models, 6 video models, 11 audio models, 2 specialist models | Gemini 4 Carbon (Rank 1 to 5) | 2 | 19 | 26 | |
| Meta | 32 | 29 language models, 1 image model, 1 video model, 1 specialist model | Muse Wonder Full (Rank 6 to 9) | 2 | 26 | 7 |
| DeepSeek | 23 | 23 language models | DeepSeek V4.1 Flash 1017 (Rank 13 to 20) | 0 | 23 | 9 |
| Moonshot AI | 12 | 12 language models | Kimi K3.1 Pro (Rank 10 to 13) | 0 | 10 | 7 |
| xAI | 16 | 12 language models, 2 image models, 2 video models | Grok 4.7 (Rank 18 to 24) | 0 | 1 | 12 |
| Alibaba | 67 | 59 language models, 3 image models, 3 video models, 2 audio models | Qwen 3.8 Max 0902 (Rank 22 to 32) | 0 | 54 | 27 |
By release date. Grey: retired models.
27 releases since January 2026, by release date.
Prices per 1M tokens for the current language models, according to Alibaba or the provider linked in each model profile.
| Current model | Input | Output | Context |
|---|---|---|---|
| Qwen 3.8 Max 0902 | $1.65 | $4.95 | 1M |
| Qwen 3.7 Max | $1.65 | $4.95 | 1M |
| Qwen 3.7 Plus | $0.28 | $1.10 | 1M |
| Qwen 3.8 27B | $0.04 | $2.50 | 262K |
| Qwen 3.7 Flash | $0.028 | $0.11 | 1M |
| Qwen 3 Coder 30B A3B | $0.07 | $0.28 | 262K |
| Qwen 3 Next 80B A3B Instruct | $0.09 | $1.10 | 262K |
| Qwen 3 30B A3B Instruct 2507 | $0.09 | $0.30 | 262K |
| Qwen 3.6 35B A3B | $0.10 | $0.95 | 262K |
| Qwen 3.5 9B | $0.10 | $0.15 | 262K |
| Qwen 3 14B | $0.10 | $0.22 | 33K |
| Qwen 2.5 7B | $0.10 | $0.20 | 131K |
| Qwen3.8-Omni-Flash | $0.11 | $0.38 | 1M |
| Qwen 3.8 Flash | $0.11 | $0.38 | 1M |
| Qwen 3 Coder Next | $0.12 | $0.80 | 262K |
| Qwen 3 30B A3B | $0.12 | $0.50 | 33K |
| Qwen 3 32B | $0.14 | $0.40 | 33K |
| Qwen 3.5 35B A3B | $0.15 | $1 | 262K |
| Qwen 3 Next 80B A3B Thinking | $0.15 | $1.20 | 262K |
| Qwen 3 235B A22B Instruct 2507 | $0.15 | $0.75 | 262K |
| Qwen 3 VL 30B A3B | $0.20 | $0.70 | 262K |
| Qwen 3 VL 235B A22B | $0.21 | $1.90 | 262K |
| Qwen 3 VL 8B | $0.25 | $0.75 | 262K |
| Qwen 3.5 122B A10B | $0.26 | $2.08 | 262K |
| Qwen 3.5 27B | $0.27 | $2.16 | 262K |
| Qwen 3 VL 30B A3B Thinking | $0.29 | $1 | 262K |
| Qwen 3.6 27B | $0.30 | $3.20 | 262K |
| Qwen 3 Coder 480B A35B | $0.30 | $1 | 262K |
| Qwen 2.5 72B | $0.36 | $0.40 | 131K |
| Qwen 3 235B A22B Thinking 2507 | $0.45 | $3.50 | 262K |
| Qwen 3.5 397B A17B | $0.55 | $3.50 | 262K |
| Qwen 2.5 Coder 32B | $0.66 | $1 | 131K |
| Qwen 2.5 VL 72B | $0.80 | $1 | 128K |
| Qwen 3.8 2.4T A95B | $2 | $6 | 262K |
| Current model | Price |
|---|---|
| Qwen Image 2.0 Pro | $0.075 per image |
| Wan 2.7 | $9 per minute |
| Wan 2.6 Text to Image | $0.03 per image |
| Wan 2.6 | $9 per minute |
| Wan 2.2 A14B | $4.80 per minute |
Monthly prices in Germany incl. VAT; “≈” is converted from US dollars. The vendor's models are often also part of other providers' plans, see the AI subscription comparison.
| Plan | Per month | Note |
|---|---|---|
| Token Plan Individual Lite | ≈ €6.39 | with credits |
| Token Plan Individual Essential | ≈ €10.65 | with credits |
| Token Plan Individual Standard | ≈ €19.17 | with credits |
| Token Plan Team Team Standard Seat | ≈ €21.30 | per seat, Team, with credits |
| Token Plan Individual Pro | ≈ €72.43 | with credits |
| Token Plan Team Team Pro Seat | ≈ €79.89 | per seat, Team, with credits |
| Token Plan Team Team Max Seat | ≈ €213.03 | per seat, Team, with credits |
The directory lists 67 Qwen models from Alibaba: 59 language models, 3 image models, 3 video models and 2 audio models. It counts every model Alibaba ever released, including retired ones; 63 of them have their own profile. 63 are currently available, the newest being Qwen3.8-Omni-Flash-Realtime, Qwen-Image-2.1 and Qwen3.8-LiveTranslate.
Qwen 3.8 Max 0902. It is at rank 22 to 32 of 95 in the overall index, which combines independent benchmarks from several sources.
Qwen3.8-Omni-Flash-Realtime, released on Sep 21, 2026. The timeline below shows every release by date.
Through the API per 1M tokens (input / output): Qwen 3.7 Flash $0.028 / $0.11 and Qwen 3.8 Max 0902 $1.65 / $4.95. The prices of all 34 current language models are in the price table. As a subscription, Qwen starts at €6.39 per month incl. VAT (Token Plan Individual Lite).
The AI subscription comparison lists no free plan for Alibaba. The cheapest plan is Token Plan Individual Lite at €6.39 per month incl. VAT.
In the overall index, no. The best OpenAI model, GPT-6 Astra (rank 4 to 8 of 95), is ahead of Qwen 3.8 Max 0902 (rank 22 to 32 of 95). Single tasks can differ.
Yes, 54 of the 67 models have open weights. Each model profile lists how much memory it needs.
Qwen 3.7 Max with 203 tokens per second according to Artificial Analysis. A typical model manages 98.
Up to 1 M tokens (≈ 1,500 A4 pages), which 6 of the current models reach, including Qwen3.8-Omni-Flash. Every model's value is in the model list.
Data as of Nov 1, 2026