Observed Signal · Feb 17, 2026 · Technical Release · Source: Hello China Tech · Impact: 4/5 · Sentiment: Positive
Alibaba Releases Qwen 3.5‑Plus with Aggressive Pricing
Alibaba released Qwen 3.5-Plus, an open-source mixed‑modality foundation model with 397 billion parameters that uses a sparse mixture‑of‑experts (MoE) design to activate roughly 17 billion parameters per inference. The model natively handles text and images, supports 201 languages and accepts video inputs up to two hours. Alibaba claims performance comparable to Google’s Gemini 3 Pro and OpenAI’s GPT-5.2 across multiple benchmarks. The company priced API input at 0.8 yuan per million tokens (≈$0.11) for requests up to 128,000 tokens, a substantially lower tier price than Gemini 3 Pro. Alibaba attributes efficiency gains to sparse MoE routing, a hybrid attention mechanism, multi‑token prediction, and an attention‑gating technique; it reports inference throughput improvements versus its Qwen 3‑Max model. Independent verification of performance and full cost comparisons remain pending, and the pricing likely includes a subsidy component tied to Alibaba Cloud hosting.
A major cloud provider released a frontier open-source foundation model with aggressive API pricing and multimodal/video capabilities; this can materially shift model-access economics, developer adoption, and infrastructure demand across AI and adjacent ad/marketing technology stacks.
Track Alibaba Cloud Signals & Market Shifts in Real-Time
Polaris7 autonomous intelligence agents track regulatory filings, primary sources, executive changes, and deal flow 24/7. Create your Polaris7 workspace to monitor these entities.
Key Takeaways & Evidence Grounding
- Alibaba released Qwen 3.5-Plus, an open-source mixed‑modality AI model with 397 billion total parameters.
- The model uses a sparse mixture-of-experts architecture that activates ~17 billion parameters per inference call (claimed <5% activation).
- Alibaba claims Qwen 3.5-Plus matches Google Gemini 3 Pro and OpenAI GPT-5.2 on a range of benchmarks (not independently verified).
- API input price set at 0.8 yuan per million tokens (~$0.11) for requests up to 128,000 tokens, roughly one‑eighteenth the input price Alibaba says Gemini 3 Pro charges at comparable volume.
- Reported engineering features include hybrid attention, multi-token prediction, and an attention‑gating mechanism; Alibaba reports 8.6x inference throughput improvements versus Qwen 3‑Max.
Connected Companies & Entities
Ontology Mapping & Concepts
Related Market Signals & Shifts
Recent verified developments and strategic activity across this market segment.
Alibaba Releases Qwen 3.5 LLM Series
Alibaba's Qwen team released the Qwen 3.5 model family, led by flagship Qwen3.5-397B-A17B and a 'Medium' tier highlighted by Qwen3.5-35B-A3B. The team also published a 'Small' series (0.8B–9B parameters) designed for on-device edge deployment. Beyond scale, Qwen 3.5 represents an architectural shift: it departs from a pure dense transformer, reimagines attention mechanisms, adopts extreme Mixture-of-Experts (MoE) sparsity, and provides native multimodal capabilities at sizes suitable for smartphones. Early benchmarks position the flagship models competitive with proprietary models such as GPT-5.2 and Claude Opus 4.5. The release signals Alibaba’s intent to control more of the deployment stack and advances open-weight model engineering in both large and edge-sized configurations.
Alibaba launches Qwen3.8-Max AI model
Alibaba unveiled Qwen3.8‑Max, a 2.4‑trillion‑parameter multimodal foundation model that uses a Mixture‑of‑Experts design (activating about 95 billion parameters per request) and supports context windows up to one million tokens. Positioned as an “always‑available” colleague, it handles text, images, documents and video and emphasizes autonomous, long‑running tasks—Alibaba reported an internal software‑engineering project completed in 10–16 days and said the model outperformed many human teams and rival models. Published Arena.AI benchmarks ranked it the top Chinese model for text tasks (though behind some Anthropic/US models) and showed comparable results to Anthropic’s Fable 5 in some tests. Alibaba plans to release model weights next week and simultaneously launched QwenWork, and its shares rose after the announcement.
Alibaba Unveils Qwen4-Max: New AI Teammate for Work, Creativity, and Discovery
Alibaba has released Qwen4-Max, its largest AI model, featuring a mixture-of-experts architecture with 2.4 trillion total parameters and 71 billion active per token. It processes images, has a one-million-token context window, and is available via QwenCloud, QwenWork, and QwenStudio. Independent rankings place it ninth on Artificial Analysis with a score of 43, behind competitors like Anthropic, Google, and OpenAI; it trails in coding but excels in legal and long-context tasks. Alibaba highlights its use as an 'AI teammate' in cloud resource allocation, manufacturing, and research, including a 48-hour autonomous experiment. Priced cost-effectively, it offers cache hits at $0.075 per million input tokens and general rates of $1.50 input and $4.50 output per million tokens. Compared to Qwen3.8-Max, it performs better on Work360 while using 35% fewer tokens, reducing average task cost by ~60%. The model is proprietary, but smaller Qwen4 versions will be open-sourced.
Receive all future signals for Alibaba Cloud & competitors in your briefing
Never miss material M&A deals, leadership changes, account wins, or product rollouts. Polaris7 monitors 20,000+ primary sources 24/7, delivering verified telemetry directly into your executive briefing.
