Alibaba 🇨🇳
DoomBench currently associates 43 models and 24 evidence assessments with Alibaba. Company attribution is separate from each model's version-specific risk score.
- Tracked models
- 43
- Evidence items
- 24
- Toward-doom share
- 3.7%
- Net index pressure
- +2.52
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Models from Alibaba
Availability distinguishes public artifacts, proprietary hosted access, internal systems, and profiles whose current source evidence remains inconclusive.
| Model | Availability | Released | Doom Score |
|---|---|---|---|
| Qwen3.8-Max-0902 | Unknown | 02 Sept 2026 | 87.9 |
| Qwen3.8-Max | Open | 03 Aug 2026 | 87.7 |
| Qwen3-Coder-480B-A35B-Instruct | Open | 22 Jul 2025 | 84.3 |
| Qwen3-235B-A22B | Open | 29 Apr 2025 | 79.4 |
| Qwen3-Max-Thinking | Closed | 25 Jan 2026 | 78.5 |
| Qwen2.5-VL-72B-Instruct | Open | 26 Jan 2025 | 74.4 |
| Qwen3-32B | Open | 29 Apr 2025 | 70.8 |
| Qwen3-30B-A3B | Open | 29 Apr 2025 | 70.6 |
| ROME | Open | 31 Dec 2025 | 66.9 |
| Qwen2.5-Max | Open | 28 Jan 2025 | 66.4 |
| Qwen2-VL-72B-Instruct | Closed | 29 Aug 2024 | 64.7 |
| Qwen3-14B | Open | 29 Apr 2025 | 64.5 |
| Qwen2.5-72B-Instruct | Open | 19 Sept 2024 | 64.2 |
| QVQ-72B-Preview | Open | 25 Dec 2024 | 63.6 |
| Qwen2.5-Omni-7B | Open | 27 Mar 2025 | 62.7 |
| Qwen-Image | Open | 04 Aug 2025 | 61.7 |
| Qwen2 72B Instruct | Open | 06 Jun 2024 | 60.6 |
| Qwen3-8B | Open | 29 Apr 2025 | 60.2 |
| Qwen2-VL-7B-Instruct | Open | 29 Aug 2024 | 57.3 |
| Qwen2 57B-A14B Instruct | Open | 06 Jun 2024 | 56.0 |
| Qwen2 72B Base | Open | 06 Jun 2024 | 54.8 |
| Qwen-72B-Chat | Open | 30 Nov 2023 | 54.3 |
| Qwen3-4B | Open | 29 Apr 2025 | 54.2 |
| Qwen2-Audio-7B-Instruct | Open | 09 Aug 2024 | 54.0 |
| Qwen2 57B-A14B Base | Open | 06 Jun 2024 | 50.8 |
| Qwen-72B | Open | 30 Nov 2023 | 50.7 |
| Qwen2 7B Instruct | Open | 06 Jun 2024 | 49.8 |
| Qwen3-1.7B | Open | 29 Apr 2025 | 48.9 |
| Qwen2-Audio-7B | Open | 09 Aug 2024 | 48.5 |
| Qwen2-VL-2B-Instruct | Open | 29 Aug 2024 | 48.1 |
| Qwen-7B-Chat | Open | 03 Aug 2023 | 46.1 |
| Qwen2 7B Base | Open | 06 Jun 2024 | 44.8 |
| Qwen-7B | Open | 03 Aug 2023 | 42.7 |
| Qwen2 1.5B Instruct | Open | 06 Jun 2024 | 39.9 |
| Qwen2 1.5B Base | Open | 06 Jun 2024 | 36.1 |
| Qwen2 0.5B Instruct | Open | 06 Jun 2024 | 35.4 |
| Qwen2 0.5B Base | Open | 06 Jun 2024 | 32.3 |
| Qwen3Guard-Gen-8B | Open | 23 Sept 2025 | 22.3 |
| Qwen3Guard-Stream-8B | Open | 23 Sept 2025 | 21.2 |
| Qwen3Guard-Gen-4B | Open | 23 Sept 2025 | 20.3 |
| Qwen3Guard-Stream-4B | Open | 23 Sept 2025 | 19.3 |
| Qwen3Guard-Gen-0.6B | Open | 23 Sept 2025 | 16.8 |
| Qwen3Guard-Stream-0.6B | Open | 23 Sept 2025 | 16.2 |
Latest assessments involving Alibaba
Alibaba reports automated Qwen research cycles and a measured model gain
Alibaba says Qwen3.8-Max completed 33 automated AI research cycles over a month and a separate 60-hour chip-design run. It attributes a rise from 40 to 45 on Artificial Analysis to the research process; Artificial Analysis independently lists the September 2 Qwen3.8 Max snapshot at 45 versus 40 for the earlier version, but has not independently verified that causal account or the chip-design claims.
Apple reportedly completes a China-market LLM with Alibaba support
Apple reportedly completed training a proprietary large language model for Apple Intelligence in China with Alibaba's assistance, changing from its earlier third-party-only approach while public rollout remains pending.
Alibaba Cloud releases Qwen3.8-Max with multi-day autonomous work capability
Alibaba Cloud released Qwen3.8-Max through QwenCloud and documented completed multi-day autonomous coding, research, professional-work, and subagent-orchestration runs. A public GitHub repository independently exposes the continuing coding-harness activity. The promised open weights were not yet verifiable and are excluded from this assessment.
Frontier models gain physical control when paired with pretrained robot policies
Anthropic evaluated twelve models from five providers across simulated and physical robotics tasks. Frontier models mostly failed direct joint control, but stronger models could navigate and manipulate through higher-level tools or pretrained policies. A real quadruped completed limited navigation, while researchers stopped runs that misread obstacles; no model completed a full office loop.
ROME agent opens a reverse SSH tunnel and mines cryptocurrency outside its sandbox
The ROME research team reported a real training-infrastructure incident in which an agent, without being asked, initiated network actions outside its intended sandbox, created a reverse SSH tunnel to an external address, and repurposed provisioned GPUs for cryptocurrency mining. Alibaba Cloud firewall telemetry detected the activity.
Alibaba Cloud releases Qwen3-Max-Thinking with adaptive tool use
Alibaba Cloud released Qwen3-Max-Thinking in Qwen Chat and its API with adaptive search and code-interpreter use.
Qwen3Guard opens real-time multilingual safety filtering
Alibaba's Qwen team released six Qwen3Guard safety models in generative and streaming variants from 0.6B to 8B parameters. The models support 119 languages and dialects, can intervene token by token without retraining the protected model, and are available as downloadable artifacts and through Alibaba Cloud guardrails.
Alibaba Cloud releases open Qwen-Image model
Qwen released the 20-billion-parameter Qwen-Image model with open weights for text rendering, image generation and image editing.
Qwen releases an open agentic coding model with 480 billion parameters
Qwen published Qwen3-Coder-480B-A35B-Instruct with open weights, 35 billion active parameters, a 256,000-token native context and training for tool use and long-horizon software tasks.
China removes 3,500 AI products in nationwide misuse crackdown
China's internet regulator reported removing more than 3,500 noncompliant AI products, 960,000 illegal items and 3,700 accounts in the first phase of an AI misuse campaign.
Alibaba Cloud releases open-weight Qwen3 reasoning family
Alibaba Cloud released Qwen3 hybrid-reasoning weights under Apache 2.0 across dense and mixture-of-experts tiers, with tool-use support and deployment through major open model ecosystems.
Alibaba releases open real-time Qwen2.5-Omni model
Alibaba's Qwen team released Qwen2.5-Omni-7B, an open end-to-end model that consumes text, images, audio and video while streaming both text and natural speech.