Alibaba
Alibaba ranks #10 of 91 providers on BenchLeader by its best model, with 54 ranked models. Its best model is Qwen3.8 Max at 63.3 on the BenchLeader Index, strongest in reasoning. Prices run from $0.060 to $3.75 per million tokens blended; Qwen3.5-4B is the cheapest ranked option. The fastest is Qwen3 Coder 30B A3B Instruct at 244 tokens per second. 48 of its 114 models ship open weights. It has shipped 48 dated releases, roughly one every 7 days; the last was 17 Sept 2026, so the next would be due between 21 Sept 2026 and 26 Sept 2026 on that pattern.
Alibaba Cloud's Qwen team ships the broadest open-weights line-up (Qwen3, Qwen3.5, coder and VL variants) plus hosted Qwen Max models.
- Provider rank
- #10
- by best model
- Best index
- 63.3
- Qwen3.8 Max (max)
- Ranked models
- 54
- 48 open weights
- Cheapest ranked
- $0.060/M
- Qwen3.5-4B
Release timeline
Each model’s best configuration by release date. The top eight are labelled; hover any point for its name.
- Alibaba54
Next release
Over the last two years Alibaba shipped 48 distinct models, one every 7 days on median. The latest arrived on 17 Sept 2026.
If the cadence holds, the next one lands between 21 Sept 2026 and 26 Sept 2026.
An estimate from past release dates only. It knows nothing about roadmaps or announcements.
Fastest ranked model: Qwen3 Coder 30B A3B Instruct at 244 tok/s.
Head to head
Alibaba vs Anthropic
Alibaba vs Google
Alibaba vs Meta
Alibaba vs OpenAI
Alibaba vs Moonshot AI
Alibaba vs SpaceXAI
Alibaba vs DeepSeek
Alibaba vs Xiaomi
Alibaba vs Zhipu AI
Alibaba vs MiniMax
Alibaba vs StepFun
Alibaba vs Sapiens AI
Alibaba vs ByteDance
Alibaba vs Motif Technologies
Alibaba vs Mistral AI
Alibaba vs MBZUAI
Alibaba vs Meituan
Alibaba vs Multiverse Computing
Alibaba vs NVIDIAAlibaba models compared
Qwen3.8 Max vs Claude Opus 5.5Qwen3.8 Max vs Gemini 4 ArgonQwen3.8 Max vs Muse Wonder FullQwen3.8 Max vs Claude Fable 5.1Qwen3.8 Max vs GPT-6 AstraQwen3.6 Max vs Claude Opus 5.5Qwen3.6 Max vs Gemini 4 ArgonQwen3.6 Max vs Muse Wonder FullQwen3.6 Max vs Claude Fable 5.1Qwen3.6 Max vs GPT-6 AstraQwen3.8 2.4T A95B vs Claude Opus 5.5Qwen3.8 2.4T A95B vs Gemini 4 ArgonQwen3.8 2.4T A95B vs Muse Wonder FullQwen3.8 2.4T A95B vs Claude Fable 5.1Qwen3.8 2.4T A95B vs GPT-6 Astra
Models
Every Alibaba model with a result, one row per model; switch to all configurations to see each reasoning effort.
114 of 114
| # | ||||||||
|---|---|---|---|---|---|---|---|---|
| 1 | Qwen3.8 Maxmax | 63.3±4 | $3.00 | 42 | 51 s | 1M | 2026-08-03 | 2026-10-23 |
| 2 | Qwen3.6 Maxmax | 62.3±3 | $2.96 | – | – | 246k | 2026-04-20 | 2026-10-23 |
| 3 | Qwen3.8 2.4T A95Bopen ↗ | 61.0±15 | $3.00 | 43 | 49 s | 1M | 2026-08-12 | 2026-08-12 |
| 4 | Qwen3.7max | 60.6±4 | $3.75 | 201 | 14 s | 1M | 2026-05-19 | 2026-10-23 |
| 5 | Qwen3.8-Flash-Nextopen ↗ | 60.3±6 | $0.230 | 52 | 41 s | 256k | 2026-08-26 | 2026-10-23 |
| 6 | Qwen3.6 Plus | 58.8±4 | $1.13 | 56 | 102 s | 1M | 2026-03-31 | 2026-10-23 |
| 7 | Qwen3.7 Plus | 58.5±6 | $0.700 | 59 | 36 s | 1M | 2026-06-02 | 2026-10-23 |
| 8 | Qwen3.5 Plusthinking | 57.3±1 | $0.258 | 54 | 1.01 s | 1M | 2026-02-16 | 2026-10-13 |
| 9 | Qwen3.8 27Bxhighopen ↗ | 56.3±6 | $1.13 | 44 | 50 s | 262k | 2026-08-14 | 2026-10-20 |
| 10 | Qwen3.5 397B A17Bthinkingopen ↗ | 55.1±8 | $1.35 | 58 | 57 s | 262k | 2026-02-13 | 2026-03-02 · stale |
| 11 | Qwen3.5-27Bthinkingopen ↗ | 54.7±8 | $0.825 | 78 | 31 s | 262k | 2026-02-24 | 2026-02-24 · stale |
| 12 | Qwen3.5-122B-A10Bopen ↗ | 54.3±6 | $1.10 | 57 | 1.29 s | 262k | 2026-02-24 | 2026-10-23 |
| 13 | Qwen3.6 27Bthinkingopen ↗ | 53.9±6 | $1.35 | 58 | 102 s | 262k | 2026-04-22 | 2026-04-22 · stale |
| 14 | Qwen3max | 53.2±7 | – | – | – | 262k | 2026-01-26 | 2026-01-26 · stale |
| 15 | Qwen3 Maxmax | 52.8±3 | $0.628 | – | – | 262k | 2025-09-05 | 2026-10-22 |
| 16 | Qwen3.5-35B-A3Bthinkingopen ↗ | 52.8±7 | $0.688 | 148 | 15 s | 262k | 2026-02-24 | 2026-02-24 · stale |
| 17 | Qwen3.6 35B A3Bthinkingopen ↗ | 52.2±7 | $0.848 | 136 | 42 s | 262k | 2026-04-14 | 2026-04-16 · stale |
| 18 | Qwen3.5 Flash | 52.0±6 | $0.094 | 87 | 0.56 s | 1M | 2026-02-25 | 2026-10-23 |
| 19 | Qwen3 32Bopen ↗ | 51.5±5 | $0.502 | 28 | 0.41 s | 131k | 2025-04-29 | 2026-10-22 |
| 20 | Qwen3.5 Omni Plus | 51.4±3 | $1.50 | 112 | 1.99 s | 256k | 2026-03-30 | 2026-03-30 · stale |
| 21 | Qwen3 VL 235B A22B Instructthinkingopen ↗ | 51.0±4 | $0.513 | 39 | 1.27 s | 131k | 2025-09-23 | 2026-10-22 |
| 22 | Qwen3 235B A22B 2507open ↗ | 50.4±6 | $0.368 | 37 | 1.34 s | 262k | 2025-07-22 | 2026-10-22 |
| 23 | Qwen2.5-Maxmax | 50.2±5 | – | – | – | 32k | 2025-01-25 | 2026-10-22 |
| 24 | Qwen3 235B A22B(Apr 2025)superseded ↗open ↗ | 50.1±4 | $0.502 | – | – | 131k | 2025-04-28 | 2026-10-22 |
| 25 | Qwen3-Next 80B A3Bthinkingopen ↗ | 49.3±4 | $0.350 | 96 | 0.81 s | 131k | 2025-09-11 | 2026-10-22 |
| 26 | Qwen3.5-9Bthinkingopen ↗ | 49.0±7 | $0.155 | 59 | 36 s | 262k | 2026-02-24 | 2026-03-02 · stale |
| 27 | Qwen-Plus | 48.5±6 | $0.158 | 52 | 0.48 s | 1M | 2025-01-25 | 2026-10-22 |
| 28 | Qwen3 30B A3B(Apr 2025)superseded ↗open ↗ | 48.3±4 | $0.215 | 118 | 0.47 s | 41k | 2025-04-29 | 2026-10-22 |
| 29 | Qwen3 Coder 480Bopen ↗ | 48.0±4 | $0.635 | 82 | 1.13 s | 262k | 2025-07-22 | 2026-10-23 |
| 30 | Qwen3 VL 32B Instructthinkingopen ↗ | 47.0±5 | $0.280 | – | – | 131k | 2025-10-21 | 2025-10-21 · stale |
| 31 | Qwen3 14Bopen ↗ | 46.9±8 | $0.251 | 80 | 0.71 s | 131k | 2025-04-29 | 2025-04-29 · stale |
| 32 | Qwen3.5-4Bthinkingopen ↗ | 46.6±5 | $0.060 | 17 | 120 s | 66k | 2026-02-24 | 2026-03-02 · stale |
| 33 | Qwen3 30B A3B 2507open ↗ | 45.6±5 | $0.150 | 184 | 0.96 s | 262k | 2025-07-30 | 2026-10-22 |
| 34 | QwQ-32Bopen ↗ | 45.4±4 | – | – | – | 131k | 2025-03-05 | 2026-10-22 |
| 35 | Qwen3.5 Omni Flash | 44.6±4 | $0.275 | 234 | 1.79 s | 256k | 2026-03-30 | 2026-03-30 · stale |
| 36 | Qwen3.6 Flash | 44.1±12 | $0.371 | 68 | 0.59 s | 1M | 2026-04-27 | 2026-04-27 · stale |
| 37 | Qwen3 VL 30B A3B Instructthinkingopen ↗ | 44.0±3 | $0.413 | 27 | 2.93 s | 131k | 2025-10-03 | 2025-10-03 · stale |
| 38 | Qwen3 Coder 30B A3B Instructopen ↗ | 43.7±3 | $0.335 | 244 | 1.07 s | 262k | 2025-07-31 | 2025-07-31 · stale |
| 39 | Qwen3 VL 8B Instructthinkingopen ↗ | 42.7±1 | $0.375 | 39 | 0.98 s | 262k | 2025-10-14 | 2025-10-14 · stale |
| 40 | Qwen2.5 72Bopen ↗ | 42.2±4 | $0.485 | 27 | 3.99 s | 33k | 2024-09-19 | 2026-10-22 |
| 41 | Qwen1 5 110Bopen | 42.2±1 | – | – | – | 32k | 2024-04-25 | 2026-10-22 |
| 42 | Qwen3 Coder Nextopen ↗ | 42.1±4 | $0.483 | 88 | 1.05 s | 262k | 2026-02-02 | 2026-02-03 · stale |
| 43 | Qwen2.5 Coder 32B Instructopen ↗ | 42.1±8 | $0.745 | 30 | 1.62 s | 131k | 2024-09-18 | 2026-10-22 |
| 44 | Qwen3 Omni 30B A3B Instructthinkingopen ↗ | 41.2±5 | $0.430 | 102 | 21 s | 66k | 2025-09-22 | 2025-09-22 · stale |
| 45 | Qwen3-4B-Instruct-2507open | 40.9±4 | – | – | – | 262k | 2025-08-06 | 2025-08-06 · stale |
| 46 | Qwen2-72Bopen ↗ | 40.9±4 | – | – | – | 131k | 2024-06-07 | 2026-10-22 |
| 47 | Qwen1.5-72B | 40.6±5 | – | – | – | – | 2024-02-04 | 2026-10-22 |
| 48 | Qwen2.5 7B Instructopen ↗ | 39.9±6 | $0.125 | 79 | 0.27 s | 33k | 2024-09-19 | 2025-11-24 · stale |
| 49 | Qwen1.5-32B | 39.7±4 | – | – | – | – | 2024-04-03 | 2026-10-22 |
| 50 | Qwen3.5 2Bthinkingopen ↗ | 39.4±3 | – | – | – | 262k | 2026-03-02 | 2026-03-02 · stale |
| 51 | Qwen3 8Bthinkingopen ↗ | 38.6±4 | $0.126 | – | – | 131k | 2025-04-28 | 2025-04-28 · stale |
| 52 | Qwen3-1.7Bthinkingopen ↗ | 37.9±5 | – | – | – | 32k | 2025-04-29 | 2025-04-28 · stale |
| 53 | Qwen3 0 6Bthinkingopen | 37.4±5 | – | – | – | 32k | 2025-04-28 | 2025-04-28 · stale |
| 54 | Qwen3.5 0.8Bthinkingopen ↗ | 36.0±8 | – | – | – | 262k | 2026-03-02 | 2026-03-02 · stale |
| 55 | Qwen Turbo | – | $0.055 | – | – | 1M | 2024-11-01 | 2024-11-18 · stale |
| 56 | Qwen-14B | – | – | – | – | – | 2023-09-24 | 2026-10-22 |
| 57 | Qwen1 5 14B | – | – | – | – | – | 2024-02-04 | 2026-10-22 |
| 58 | Qwen1 5 4B | – | – | – | – | – | – | 2026-10-22 |
| 59 | Qwen1 5 7B | – | – | – | – | – | 2024-02-04 | 2026-10-22 |
| 60 | Qwen2.5 32B Instructopen ↗ | – | – | – | – | 128k | 2024-09-17 | 2024-09-19 · stale |
| 61 | Qwen2.5 Plus 1127 | – | – | – | – | – | – | 2026-10-22 |
| 62 | Qwen3.5 Maxmax | – | – | – | – | – | – | 2026-10-22 |
| 63 | Qwen2-VL-72B | – | – | – | – | – | 2024-08-29 | 2026-10-19 |
| 64 | Qwen2.5 VL 72B Instructopen ↗ | – | $0.850 | 30 | 1.09 s | 128k | 2024-09-19 | 2026-10-19 |
| 65 | Qwen3 4Bthinkingsuperseded ↗open ↗ | – | – | – | – | 32k | 2025-04-29 | 2025-04-28 · stale |
| 66 | Qwen3.7 Flashno reasoning | – | $0.052 | 72 | 0.71 s | 1M | 2026-07-27 | 2026-07-27 |
| 67 | Qwen Qwen2 5 72B Turbo | – | – | – | – | – | – | 2026-10-13 |
| 68 | Qwen-VLmax | – | $0.316 | – | – | 131k | 2024-01-18 | 2026-10-19 |
| 69 | Qwen Qwen2 5 7B Turbo | – | – | – | – | – | – | 2026-10-13 |
| 70 | Qwen-VL Plus | – | $0.158 | – | – | 131k | 2024-01-25 | 2023-12-15 · stale |
| 71 | Qwen-VL-Max-1119max | – | – | – | – | – | – | 2026-10-19 |
| 72 | Qwen2-VL-7B | – | – | – | – | – | 2024-08-29 | 2026-10-19 |
| 73 | Qwen2.5-Coder 7B Instructopen ↗ | – | – | – | – | 131k | 2024-09-18 | 2024-09-19 · stale |
| 74 | Qwen2.5-VL 7B Instructopen ↗ | – | – | – | – | – | 2024-09 | 2025-02-19 · stale |
| 75 | Qwen2.5-VL-32B | – | – | – | – | – | – | 2026-10-19 |
| 76 | Qwen3 235B 2507thinking | – | – | – | – | – | – | – |
| 77 | Qwen3 235B A22B (Parasail) | – | – | – | – | – | 2025-07-25 | 2025-09-01 · stale |
| 78 | Qwen3 32B Fastthinking | – | – | – | – | – | – | – |
| 79 | Qwen3.8 Flashopen ↗ | – | $0.189 | 57 | 1.04 s | 1M | 2026-08-26 | 2026-10-14 |
| 80 | Qwq 32B Plus Qwen 2.5 Coder | – | – | – | – | – | – | 2025-03-07 · stale |
| 81 | QwQ Plus | – | $0.316 | – | – | 131k | 2025-04-08 | 2025-04-08 · stale |
| 82 | CodeQwen1.5-7B | – | – | – | – | – | 2024-04-15 | – |
| 83 | mPLUG-Owl3-7B-241101 | – | – | – | – | – | 2024-11-26 | – |
| 84 | Qwen Deep Research | – | $11.65 | – | – | 1M | 2024-01 | – |
| 85 | Qwen Doc Turbo | – | $0.101 | – | – | 131k | 2024-01 | – |
| 86 | Qwen Flash | – | $0.071 | – | – | 1M | 2025-07-28 | – |
| 87 | Qwen Long | – | $0.126 | – | – | 10M | 2025-01-25 | – |
| 88 | Qwen Math Plus | – | $0.861 | – | – | 4k | 2024-08-16 | – |
| 89 | Qwen Math Turbo | – | $0.431 | – | – | 4k | 2024-09-19 | – |
| 90 | Qwen Plus Character | – | $0.158 | – | – | 33k | 2024-01 | – |
| 91 | Qwen Plus Character (Japanese) | – | $0.725 | – | – | 8k | 2024-01 | – |
| 92 | Qwen-1_8B | – | – | – | – | – | 2023-11-30 | – |
| 93 | Qwen-7B | – | – | – | – | – | 2023-09-28 | – |
| 94 | Qwen-MT Plus | – | $0.388 | – | – | 16k | 2025-01 | – |
| 95 | Qwen-MT Turbo | – | $0.146 | – | – | 16k | 2025-01 | – |
| 96 | Qwen-Omni Turbo | – | $0.101 | – | – | 33k | 2025-01-19 | – |
| 97 | Qwen-VL OCR | – | $0.050 | – | – | 34k | 2024-10-28 | – |
| 98 | Qwen2.5 14B Instructopen ↗ | – | – | – | – | – | 2024-09-19 | – |
| 99 | Qwen2.5 Coder | – | – | – | – | – | – | – |
| 100 | Qwen2.5 Coder 1 5B | – | – | – | – | – | 2024-09-18 | – |
Latest news
- China’s Free A.I. Models Are Thrown Into UncertaintyThe New York Times · 21 Oct 2026
- Alibaba deepens AI push with new chip, bigger model; shares jump 5%Reuters · 21 Sept 2026
- US Says Alibaba, DeepSeek Have ‘Systematically’ Siphoned AI ModelsBloomberg · 9 Sept 2026
- Alibaba open-sources AI model that can detect cancer and nearly 150 conditionsSouth China Morning Post · 18 Sept 2026
- Alibaba AI model writes more hackable code for US contractorsTechGig · 22 Oct 2026
- Here’s How China-Based Alibaba’s New AI Model Compares With ChatGPT And FableForbes · 3 Aug 2026
- Chinese AI Models Dominate South Korea's Development Ecosystem조선일보 · 20 Oct 2026
- Alibaba AI models cleared for use in Apple Intelligence in China just in time for the iPhone Duo launchEngadget · 14 Oct 2026