Organization

Alibaba

Tracked models

122

Largest training run

1.5 x 10^25 FLOP

Data-center compute

31.9K H100e

122 tracked AI models list Alibaba as a developer. The largest is Qwen3-Max, at 1.5 x 10^25 FLOP (September 2025). The most recent tracked release is from May 2026.

It owns 1 tracked AI data center, holding about 31.9K H100e and drawing 169 MW. It operates 7 tracked GPU clusters.

Models (80+)

Qwen3-Max1.5 x 10^25 FLOP Qwen2.5-VL-72B9.6 x 10^24 FLOP Qwen2.5 Instruct (72B)7.9 x 10^24 FLOP Qwen2.5-72B7.8 x 10^24 FLOP Qwen3-32B7.1 x 10^24 FLOP Qwen3-235B-A22B (Jul 2025)4.8 x 10^24 FLOP Qwen3-235B-A22B-Thinking (Jul 2025)4.8 x 10^24 FLOP Qwen3-235B-A22B4.8 x 10^24 FLOP QwQ-32B3.5 x 10^24 FLOP Qwen2.5-32B3.5 x 10^24 FLOP Qwen3-14B3.2 x 10^24 FLOP Qwen2-72B3 x 10^24 FLOP Qwen3-8B1.8 x 10^24 FLOP Qwen2.5-14B1.6 x 10^24 FLOP Qwen3-Coder-480B-A35B1.6 x 10^24 FLOP Qwen1.5-72B1.3 x 10^24 FLOP Qwen-72B1.3 x 10^24 FLOP Qwen2.5-Coder (32B)1.1 x 10^24 FLOP Qwen2.5-VL-7B9.9 x 10^23 FLOP Qwen3-4B8.6 x 10^23 FLOP Qwen2.5-Math-7B-Base8.6 x 10^23 FLOP Qwen2.5-7B8.2 x 10^23 FLOP AgentFounder-30B6.5 x 10^23 FLOP Qwen3-30B-A3B6.5 x 10^23 FLOP Qwen2-VL-72B6 x 10^23 FLOP Wan 2.2 14B T2V4.2 x 10^23 FLOP Qwen2.5-VL-3B4.1 x 10^23 FLOP Qwen2-57B-A14B3.8 x 10^23 FLOP Qwen3-1.7B3.7 x 10^23 FLOP Qwen1.5-14B3.4 x 10^23 FLOP Qwen2.5-3B3.3 x 10^23 FLOP Qwen2-7B2.9 x 10^23 FLOP Qwen3-Next-80B-A3B2.7 x 10^23 FLOP Qwen2.5-Coder (7B)2.5 x 10^23 FLOP Wan 2.1 14B I2V2.5 x 10^23 FLOP Qwen-14B2.5 x 10^23 FLOP Qwen2.5-Math-1.5B1.8 x 10^23 FLOP Ovis-7B1.7 x 10^23 FLOP Qwen1.5-7B1.7 x 10^23 FLOP Qwen2.5-1.5B1.7 x 10^23 FLOP Qwen3-0.6B1.3 x 10^23 FLOP Qwen-7B10^23 FLOP Qwen2-VL-7B6.7 x 10^22 FLOP Qwen2-1.5B6.3 x 10^22 FLOP Qwen2.5-Coder (1.5B)5.1 x 10^22 FLOP Qwen3-Omni-Flash3.6 x 10^22 FLOP Qwen3-Omni-30B-A3B3.6 x 10^22 FLOP Qwen2-0.5B3.6 x 10^22 FLOP PLUG3.6 x 10^22 FLOP Qwen2-VL-2B1.7 x 10^22 FLOP M6-10T5.5 x 10^21 FLOP M6-T5.5 x 10^21 FLOP gte-modernbert3.7 x 10^21 FLOP AR-LDM5.1 x 10^20 FLOP ONE-PEACE1.8 x 10^20 FLOP Qwen 3.7 MaxMay 2026 Qwen3.7-MaxMay 2026 Qwen 3.6 PlusApril 2026 Qwen3.5-Omni-PlusMarch 2026 Qwen3.5-Omni-FlashMarch 2026 Qwen 3.5 Flash (hosted 35B-A3B)February 2026 Qwen3.5-27BFebruary 2026 Qwen3.5-9BFebruary 2026 Qwen3.5-4BFebruary 2026 Qwen3.5-2BFebruary 2026 Qwen3.5-0.8BFebruary 2026 Qwen3.5-122B-A10BFebruary 2026 Qwen 3.5 Plus (hosted 397B-A17B)February 2026 Qwen3.5 397B-A17BFebruary 2026 Qwen3-Coder-NextFebruary 2026 Qwen3-Max-ThinkingJanuary 2026 Tongyi DeepResearchOctober 2025 Wan 2.5September 2025 Wan 2.2 14B S2VAugust 2025 Qwen Image EditAugust 2025 Ovis2.5 9BAugust 2025 Ovis2.5 2BAugust 2025 Qwen ImageAugust 2025 Wan 2.2 14B I2VJuly 2025 Qwen3 EmbeddingJune 2025

Data centers (1)

Alibaba Zhangbei31.9K H100e

GPU clusters (7)

Alibaba Cloud in Thailand Phase 1 Alibaba Cloud in Thailand Phase 2 Alibaba Cloud in South Korea Phase 2 Alibaba Cloud in Mexico Alibaba Cloud in Philippines Phase 1 Alibaba Cloud in South Korea Phase 1 Alibaba Cloud in Malaysia Phase 2

SourceEpoch AI, 'AI Models'. Published online at epoch.ai. Retrieved 2026-07-29 from https://epoch.ai/data/ai-models. Licensed under CC BY 4.0.