Alibaba

company↑ rising

AlibabaAlibaba Group Holding Ltd.

Alibaba Cloud + DAMO Academy. Qwen open-source models compete with Llama, Mistral. Largest cloud in Asia. Expanding Qwen-VL, Qwen-Audio, Qwen-Agent.

🤖Agent's take · Momentum3d ago · graph-walked

Alibaba, through DAMO Academy and Alibaba Cloud, is executing the most aggressive open-source model release cadence in AI. Since Qwen2, it has shipped Qwen3-235B, Qwen3-VL-4B-Instruct, Qwen3.5-35B, Qwen 3.7-Max, Qwen 3.6 Plus, Qwen3-30B-A3B, Qwen 3.5 397B, Qwen-Image, Qwen3.5-Omni, and Qwen 3.5 Medium — covering vision, audio, agent, and reasoning modalities. This breadth directly challenges Llama and Mistral in the open-weight arena. Yet the graph reveals a critical tension: Alibaba competes with DeepSeek while also holding an investment in it. It also invests in Moonshot AI, signaling a hedge against domestic rivals. As Asia's largest cloud provider, Alibaba can distribute these models at scale, but its competition with Nvidia, Google, Tencent, and ByteDance means it must fend off both hyperscalers and AI-native startups. The question: can Alibaba's open-source blitz maintain developer mindshare as DeepSeek and ByteDance accelerate their own releases?

·Shipped 10+ Qwen models across vision, audio, agent, and reasoning since Qwen2.
·Competes with Llama, Mistral, Nvidia, Google, Tencent, ByteDance, and DeepSeek.
·Invested in both DeepSeek and Moonshot AI, hedging against domestic AI startups.
·Leverages Alibaba Cloud, Asia's largest cloud, for model distribution.
·Qwen3.5-Omni and Qwen-Image extend multimodal coverage beyond text.

73Total Mentions

+0.31Sentiment (Positive)

+1.4%Velocity (7d)

Public (BABA)Total Funding

View subgraph

Hangzhou, China200,000 employeesFirst seen: Feb 16, 2026Last active: 2d agoWikipedia

Signal Radar

Five-axis snapshot of this entity's footprint

live

Loading radar…

Mentions × Lab Attention

Weekly mentions (solid) and average article relevance (dotted)

mentionsrelevance

Loading timeline…

Timeline

Research MilestoneMay 25, 2026
Alibaba and Nanjing University published paper claiming 9.36X speedup for million-token prefill
View source
speedup:
9.36X
baseline:
FlashAttention-2
PartnershipApr 23, 2026
Launched first external partnership for Qwen AI app with China Eastern Airlines for flight booking integration
View source
partner:
China Eastern Airlines
capability:
end-to-end flight management via chat
Product LaunchApr 22, 2026
Released Qwen3.6-27B, a dense 27B parameter model that surpasses its previous 397B MoE model on coding benchmarks.
View source
model size:
27B parameters
file size:
55.6GB (16.8GB quantized)
context window:
65536 tokens
Research MilestoneApr 20, 2026
Researchers developed DCW (Diffusion Correction in Wavelet domain), a method to fix SNR-t misalignment bias in diffusion models, improving performance for models like FLUX and EDM.
View source
Product LaunchApr 19, 2026
Moved Qwen 3.6 Plus model to API-only access, adopting a paid frontier model strategy.
View source
Product LaunchApr 17, 2026
Released Qwen 3.6, a free, open-weights AI model for local deployment and coding
View source
Product LaunchApr 16, 2026
Released Qwen3.6-35B-A3B, a sparse MoE model with 35B total and 3B active parameters
View source
model:
Qwen3.6-35B-A3B
architecture:
sparse Mixture-of-Experts
benchmark:
SWE-bench Verified: 73.4%
Product LaunchApr 15, 2026
Amap division launched three embodied AI models (ABot-World, ABot-M0, ABot-N0) claiming top benchmarks.
View source
Research MilestoneApr 9, 2026
Published research paper on AI moving beyond text to multi-modal capabilities.
View source
Research MilestoneApr 9, 2026
Published VulnSage, a multi-agent LLM framework that generated 146 zero-day exploits.
View source
vulnerabilities found:
146
performance gain:
34.64% more successful exploits
Product LaunchApr 5, 2026
Launch of Qwen 3.6-Plus AI model
View source
Product LaunchApr 3, 2026
Alibaba launches Qwen3.6-Plus with 1M-token context
Product LaunchMar 31, 2026
Launched Qwen 3.5 Omni model with advanced voice AI and strategic messaging targeting the Western market.
View source
Product LaunchMar 30, 2026
Released Qwen3.5-Omni multimodal model with script-level captioning and audio-visual vibe coding
View source
Research MilestoneMar 25, 2026
Proposed and deployed the KARMA framework for personalized search on Taobao
View source
metric improvement:
increased item clicks by +0.5%
Product LaunchMar 25, 2026
Aggressively releasing open-source AI models (Qwen) as part of China's ecosystem development strategy.
View source
product:
Qwen models
license:
Open-source
Product LaunchMar 24, 2026
Launched the XuanTie C950 CPU, a high-performance RISC-V processor with native LLM support
View source
Research MilestoneMar 24, 2026
Achieved record SPECint2006 score (>70) for RISC-V architecture with XuanTie C950
View source
Product LaunchMar 23, 2026
Announced at ModelScope DevCon that more open-source Qwen models will be released
View source
Research MilestoneMar 23, 2026
Researchers propose AIGQ, an end-to-end generative framework for pre-search query recommendations
View source
application:
Taobao HintQ system

Relationships

Developed

→
Qwen 3.5 Medium
ai model✓ corroborated7 sources30% conf.
→
Qwen 3.6 Plus
ai model✓ corroborated6 sources30% conf.
→
Qwen 2.5
ai model✓ corroborated5 sources13% conf.
→
Qwen3.5-Omni
ai model✓ corroborated4 sources30% conf.
→
Qwen 3.5 397B
ai model✓ corroborated3 sources49% conf.
→
Qwen 3.6
ai model✓ corroborated3 sources30% conf.
→
Qwen 3.5 4B
ai model✓ corroborated3 sources13% conf.
→
OpenSandbox
product✓ corroborated3 sources13% conf.
→
Qwen3.6-27B
ai model✓ corroborated2 sources44% conf.
→
Qwen 3.5 397B-A17B
ai model✓ corroborated2 sources30% conf.
→
Qwen3-30B-A3B
ai model✓ corroborated2 sources30% conf.
→
Qwen2.5-Coder-7B-Instruct
ai model✓ corroborated2 sources30% conf.
→
Qwen3.5
ai model✓ corroborated2 sources13% conf.
→
Taobao
product✓ corroborated2 sources13% conf.
→
Qwen3
ai model✓ corroborated2 sources13% conf.
→
Qwen 3.7-Max
ai model1 source100% conf.
→
Qwen 2.5 72B
ai model1 source100% conf.
→
Qwen3-235B
ai model1 mention90% conf.
→
Qwen2
ai model1 mention90% conf.
→
Qwen3-VL-4B-Instruct
ai model1 mention90% conf.
→
Qwen3.5–35B
ai model1 mention90% conf.
→
Qwen 3.7
ai model1 source90% conf.
→
Amap
product1 source30% conf.
→
Qwen-Image
ai model1 source30% conf.
→
Qwen-Scope
product1 source30% conf.
→
DCW
technology1 source30% conf.

Competes With

←
Anthropic
company✓ corroborated5 mentionsCompare13% conf.
←
OpenAI
company✓ corroborated5 mentionsCompare13% conf.
→
OpenAI
company✓ corroborated3 sources Compare30% conf.
→
Baidu
company✓ corroborated3 sources Compare13% conf.
→
Tencent
company✓ corroborated3 sources Compare13% conf.
←
Baidu
company✓ corroborated3 sources Compare13% conf.
←
Tencent
company✓ corroborated3 sources Compare13% conf.
→
Anthropic
company✓ corroborated2 sources Compare30% conf.
→
Meta
company✓ corroborated2 sources Compare30% conf.
→
DeepSeek
company1 source Compare60% conf.
→
Nvidia
company1 source Compare30% conf.
→
Google
company1 source Compare30% conf.
→
Mistral AI
company1 source Compare30% conf.
→
ByteDance
company1 source Compare30% conf.
←
Meta
company1 source Compare30% conf.
←
ByteDance
company1 source Compare30% conf.
←
Mistral AI
company1 source Compare30% conf.
←
Google
company1 source Compare30% conf.
←
Nvidia
company1 source Compare30% conf.

Invested

→
Moonshot AI
company✓ corroborated3 sources93% conf.
→
DeepSeek
company1 source30% conf.

Partnered

→
Nanjing University
organization1 source95% conf.
→
China Eastern Airlines
company1 source30% conf.
←
China Eastern Airlines
company1 source30% conf.

Uses

←
OpenRouter
company1 source90% conf.

Frequently appears with

Entities that show up in the same articles — shared coverage, not a stated relationship.

Predictions

pendingquarterMay 27, 2026
Alibaba will repackage Qwen for developer workflows, not chat
Within the next quarter, Alibaba will launch or repackage Qwen around a developer workflow surface — coding, agent orchestration, or cloud-native tooling — rather than positioning it primarily as a general chatbot. The key signal will be a productized workflow layer or pricing path aimed at builders, because Alibaba is increasingly competing with OpenAI, Anthropic, and Google on multiple fronts.
35%
incorrectmonthApr 7, 2026
Alibaba announces Qwen 4.0 with OpenSandbox agent platform integration at their Cloud Summit in June 2026
Alibaba will launch Qwen 4.0 at their annual Cloud Summit (typically June) featuring native integration with OpenSandbox for building and deploying AI agents, with specific benchmarks showing superior performance on Chinese e-commerce and coding tasks compared to GPT-4o and Claude 3.5.
80%
correctmonthFeb 28, 2026
Alibaba announces major AI model focused on Asian markets
Alibaba will announce a new, major AI model or a significant partnership focused on Asian markets within the next 60 days. This is an attempt to regain momentum after declining velocity (0.4x) and capitalize on its positive sentiment shift (+0.47).
85%

AI Discoveries

observationactive3d ago
Lifecycle: Alibaba
Alibaba is in 'surging' phase (1 mentions/3d, 1/14d, 72 total)
90% confidence
observationactiveJun 8, 2026
Silence anomaly: Alibaba
Alibaba (company) has 71 total mentions but hasn't appeared in any article for 14 days. Previously active entity going quiet — may indicate strategic shift, acquisition, or pivoting away from public discourse.
70% confidence
hypothesisactiveApr 7, 2026
H: Alibaba will announce a strategic partnership with a major non-Chinese cloud provider (AWS, Google C
Alibaba will announce a strategic partnership with a major non-Chinese cloud provider (AWS, Google Cloud, or Azure) to offer Qwen models as a managed service outside China within Q3 2026.
75% confidence
hypothesisactiveApr 7, 2026
H: Alibaba's next major Qwen release (Qwen 4.0) will include native integration with their OpenSandbox
Alibaba's next major Qwen release (Qwen 4.0) will include native integration with their OpenSandbox platform, creating a closed-loop AI agent development environment that directly competes with OpenAI's GPTs and Anthropic's Claude Desktop.
85% confidence

Sentiment History

6-W176-W206-W24

Positive sentiment

Negative sentiment

Range: -1 to +1

Week	Avg Sentiment	Mentions
2026-W17	0.65	2
2026-W18	0.60	1
2026-W19	0.20	1
2026-W20	0.25	2
2026-W21	0.43	3
2026-W22	0.40	1
2026-W24	0.25	2

Alibaba

Signal Radar

Mentions × Lab Attention

Timeline

Relationships

Developed

Competes With

Invested

Partnered

Uses

Frequently appears with

Recent Articles

Rio 3.5 Open 397B: City Gov't Model Tops Open-Source Leaderboard

Moonshot AI, State Bank Launch First AI-Native Credit Card in China

Alibaba + Nanjing Univ Claim 9.36X Faster Million-Token Prefill vs FlashAttention-2

Qwen 3.7-Max Agentic Coding Demo Shows Frontier-Level UI Replication

Moonshot AI's Kimi WebBridge Lets Agent Use Your Logged-In Sessions

Predictions

Alibaba will repackage Qwen for developer workflows, not chat

Alibaba announces Qwen 4.0 with OpenSandbox agent platform integration at their Cloud Summit in June 2026

Alibaba announces major AI model focused on Asian markets

AI Discoveries

Lifecycle: Alibaba

Silence anomaly: Alibaba

H: Alibaba will announce a strategic partnership with a major non-Chinese cloud provider (AWS, Google C

H: Alibaba's next major Qwen release (Qwen 4.0) will include native integration with their OpenSandbox

Sentiment History