Text and chat · Alibaba (Qwen)
Qwen Max, Plus and Flash
Alibaba's hosted Qwen models, with 1M context and image and video input. Flash is cheap and fast. Anonymized.
Anonymized Low costWhat sets it apart
These are Alibaba’s hosted Qwen models, with a 1M token window from Plus and Max 3.7 on. Qwen 3.8 Max reaches 45 on the Artificial Analysis index at $2 / $6 per million tokens, while Plus tiers read images and video for $0.40 to $0.50 per million input tokens.
- Plus tiers read images and video
- Max 3.8 scores 45 on the index
- Plus at $0.40 per million input
Strengths and limits
Where it shines
- Reads images and video
- 1M token context on the 3.6 to 3.8 versions
- Qwen 3.8 Flash costs very little per token
Think twice when
- Anonymized, not private: Alibaba Cloud's retention policy applies
- For private mode, use the open Qwen models instead
- 1M token context (3.5 397B: 128K)
- Input: text, image, video
- Tool calling
Use cases
Analyzing a screen recording
Watch this screen recording of the bug and write the steps to reproduce it, with the exact menu names.
Qwen Plus tiers accept video input at $0.40 per million input tokens.
Complex multi-step reasoning
Plan a three-city trip with a train budget of 400 euros, two rest days and one museum day each. Show the cost per leg.
Qwen3.8 Max is the strongest hosted Qwen, 45 on the index.
Against its rivals
- GLMGLM 5.3 matches Qwen3.8 Max at 45 and publishes its weights.#13 of 90 on Artificial Analysis Intelligence Index (45 points)
- GeminiGemini 3.8 Flash scores 41 and is much faster.#5 of 90 on Artificial Analysis Intelligence Index (53 points)
- GrokGrok 4.7 costs the same per token ($2 / $6) and scores 46.#12 of 90 on Artificial Analysis Intelligence Index (46 points)
Version history
Each version, dated, with what it changed over the one before.
Hollow points are earlier versions no longer offered in Sub Rosa.
See the numbers
| Version | Released | points |
|---|---|---|
| Qwen3.5 397B A17B | 2026-02-16 | 18 |
| Qwen3.6 Plus | 2026-04 | 27 |
| Qwen3.7 Max | 2026-05 | 29 |
| Qwen3.7 Plus | 2026-06-01 | 25 |
| Qwen3.8 Max | 2026-09 | 45 |
- Qwen3.5 397B A17B
- Open-weights 397B mixture of experts (17B active) under Apache 2.0, 262K context, reads images.
- Qwen3.6 Plus
- Hosted proprietary tier with a 1M context, accepting text, image and video, at $0.50 / $3 per million tokens.
- Qwen3.7 Max
- Top hosted tier, text only, 1M context, $2.50 / $7.50 per million tokens, scoring 29 on the index.
- Qwen3.7 Plus
- Plus tier cut to $0.40 / $1.60 per million tokens, with image and video input.
- Qwen3.8 Flash
- Flash tier of the 3.8 generation. Artificial Analysis lists a related open-weights Qwen3.8-Flash-Next (180B, 6B active) scoring 40 on the index.
- Qwen3.8 MaxLatest
- Index score of 45 against 29 for Qwen3.7 Max, with image input added and a 984K window; $2 / $6 per million tokens.
Benchmarks
How to read these scores: what a benchmark tells you.
Artificial Analysis Intelligence Index: #14 of 90 models measured in the catalog. One score averaging about ten independent tests of reasoning, knowledge, math, coding and agent work, all run the same way for every model. Higher is better. Each model is listed at its highest reasoning effort. Two or three points apart is a close call; ten points is a different class of model.
See the numbers
| Model | points |
|---|---|
| 1. Claude Opus 5.5 | 58 |
| 2. Claude Sonnet 5.5 | 56 |
| 3. Claude Fable 5.1 | 53 |
| 4. GPT-6 Astra | 53 |
| 5. Gemini 4 Argon | 53 |
| 6. GPT-6.1 Sol | 52 |
| 7. Muse Wonder | 52 |
| 8. Claude Opus 5 | 51 |
| 9. Claude Fable 5 | 50 |
| 10. GPT-6 Sol | 48 |
| 14. Qwen 3.8 Max | 45 |
| 43. Qwen 3.7 Max | 29 |
| 47. Qwen 3.6 Plus Uncensored | 27 |
All 6 models in this family
| Model | Use | Privacy | Price |
|---|---|---|---|
Qwen 3.8 Flashqwen-3-8-flash | Text | Anonymized | $0.22 per million tokens written$0.000154 a page written |
Qwen 3.8 Maxqwen-3-8-max | Text | Anonymized | $3.372 per million tokens written$0.00236 a page written |
Qwen 3.7 Maxqwen-3-7-max | Text | Anonymized | $3.619 per million tokens written$0.00253 a page written |
Qwen 3.7 Plusqwen-3-7-plus | Text | Anonymized | $0.899 per million tokens written$0.000629 a page written |
Qwen 3.6 Plus Uncensoredqwen-3-6-plus | Text | Anonymized | $1.686 per million tokens written$0.00118 a page written |
Qwen 3.5 397Bqwen3-5-397b-a17b | Text | Anonymized | $2.023 per million tokens written$0.00142 a page written |
Specifications
| Specification | Qwen3.8 Max | Qwen3.7 Plus | Qwen3.7 Max | Qwen3.5 397B A17B |
|---|---|---|---|---|
| Released | September 2026 | June 1, 2026 | May 2026 | February 16, 2026 |
| Parameters | Not disclosed | Not disclosed | Not disclosed | 397B total, 17B active |
| Weights | Closed | Closed | Closed | Open, Apache 2.0 |
| Accepts | text, images | text, images, video | text | text, images |
| Produces | text | text | text | text |
| Context | 1,000K tokens | 1,000K tokens | 1,000K tokens | 128K tokens |
| Abilities | Reads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for code | Reads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for code | Reads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for code | Reads images, Uses Sub Rosa’s tools, Thinks before answering, Tuned for code |
| Price | $3.372 per million tokens written | $0.899 per million tokens written | $3.619 per million tokens written | $2.023 per million tokens written |
| Privacy | Anonymized | Anonymized | Anonymized | Anonymized |
Prices, context and lengths from the live catalog on October 16, 2026; parameters and licenses from the maker.
Sources
Everything on this page comes from these pages, read on October 16, 2026 or as dated on each score.
Show the 9 sources
- artificialanalysis.ai/models/qwen3-8-max
- artificialanalysis.ai/models/qwen3-8-flash-next
- artificialanalysis.ai/models/qwen3-7-max
- artificialanalysis.ai/models/qwen3-7-plus
- artificialanalysis.ai/models/qwen3-6-plus
- artificialanalysis.ai/models/qwen3-5-397b-a17b
- artificialanalysis.ai/leaderboards/models
- datanorth.ai/news/alibaba-releases-qwen3-8-max
- alibabacloud.com/help/en/model-studio/models