In this article
- A hybrid architecture reduces compute costs for long inputs
- Efficiency gains lower the cost of each task by more than half
- Qwen4-Max ranks second among Chinese models at a third of the predecessor’s cost
- Live Delivery returns control of long-running agents to users
- Alibaba’s models now run lab experiments autonomously
- Qwen4-Max aims to build playable games from a single prompt
- Alibaba says the model writes safer code and offers no real CBRN uplift
Alibaba launches Qwen4-Max with a cheaper architecture and lower prices than its predecessor
Alibaba has released Qwen4-Max, a flagship model featuring an updated design that reduces compute needs for long documents and runs at roughly one-third the cost per task of the previous generation. The system ranks second on the Artificial Analysis Intelligence Index behind Kimi K3.1 Pro while maintaining a price significantly lower than that competitor.
The model targets professional work in industry, science, and finance. It combines a 2.4 trillion parameter count with an active set of 71 billion parameters per token and supports a one-million-token context window for processing images natively.
A hybrid architecture reduces compute costs for long inputs
The release introduces a revamped architecture for the Qwen4 series. This design combines two methods to handle context more efficiently. Three layers compress prior information into a condensed state, and a fourth layer selects only the relevant parts. This approach cuts compute costs because the model does not apply equal attention to every element of the context.
The system adds four parallel connections between model layers with dynamic gates to stabilise information flow. An N-gram embedding layer containing 328 billion parameters supplies common token sequences as pre-built blocks. Alibaba stated that training the 2.4-trillion-parameter configuration ran stably with no sudden spikes in training loss.
Internal evaluations measure quality in bits per byte across knowledge, code, question answering, and STEM. Qwen4-Max-Base scored highest among five checkpoints. It improved by 5 to 13 percent over Qwen3.8-Max-Base while activating 25 percent fewer parameters per token.
Efficiency gains lower the cost of each task by more than half
Work360 is Alibaba’s benchmark for industrial tasks across 721 fields. On this metric, Qwen4-Max uses roughly a quarter to a third fewer tokens than its predecessor to solve the same problems. On Terminal-Bench 4.0, token use drops by nearly half.
Alibaba reports the average cost per Work360 task falls 62 percent, from $4.43 to $1.67. The benchmark score climbs from 47.74 to 52.69 at the same time.
Results on external benchmarks show mixed performance. On long-horizon agent tasks like Agents’ Last Exam v1.1, Qwen4-Max scores 46.7 points, the best result in the comparison group. It trails Opus 5.5 and the GPT models on Humanity’s Last Exam. On HealthBench Professional, it falls behind the GPT models but edges past Opus 5.5.
Qwen4-Max ranks second among Chinese models at a third of the predecessor’s cost
At its highest reasoning setting, Qwen4-Max scores 43 points on the Artificial Analysis Intelligence Index. That is nine more than Qwen3.8-Max and three behind Kimi K3.1 Pro, the top Chinese model at 46 points. The biggest gains over the predecessor come on HLE-Diamond (up 21 points to 45 percent), AA-Omniscience (up 18), and AutomationBench-AA (up 17).
Artificial Analysis puts the cost of Qwen4-Max at $3.27 per Intelligence Index task. That is about a third of Qwen3.8-Max’s $10.02 and roughly 11 percent less than Kimi K3.1 Pro’s $3.68. The savings come from three sources: token prices are 25 percent lower than the predecessor’s, cache hits cost 70 percent less, and the model burns through fewer tokens. Qwen4-Max needs about 105,000 output tokens per benchmark task, compared with 171,000 for Qwen3.8-Max.
Tencent’s Hy4 is much cheaper at $1.03 per task but scores eight points lower on the index. On Terminal-Bench 4.0, Qwen4-Max scores 48 percent, behind Kimi K3.1 Pro at 67 percent. On Cognition’s coding benchmark FrontierCode 2.0, it ties Kimi at 26 percent and trails Tencent Hy4 at 31 percent.
Through its own API, Alibaba charges $1.50 per million input tokens and $4.50 per million output tokens. Cache hits cost $0.075.
Live Delivery returns control of long-running agents to users
Alibaba builds this release around agents that handle long tasks autonomously. If users only see results at the end, they have almost no way to fix mistakes. Alibaba’s answer is “Live Delivery.” The model delivers a usable first version early and builds it out in later passes. Users can step in and change course at any point. Alibaba has open-sourced the accompanying Qwen-Live-Delivery-Plugin.
Alibaba cites several partner deployments. Automotive partners SANF and Pacific Union reportedly used Qwen4-Max to cut manual programming time for a single workpiece from about 475 minutes to 24. In finance, the model taps 90 MCP tools to pull market data from multiple sources. It delivers the results as verifiable Excel and PowerPoint files, and every number traces back to its original source.
Alibaba also uses the model internally to run its own cloud infrastructure more efficiently. According to the company, Qwen4-Max worked with the in-house solver MindOpt to cut the cost of allocating compute jobs for data annotation by an average of 10.42 percent. When distributing CPU tasks in data centers, it found optimal solutions more than 14 times faster than the comparison method.
Alibaba’s models now run lab experiments autonomously
Alibaba also shows the Qwen4 generation at work in physical labs. Paired with Mira, a system from startup Deep Principle, the model ran autonomously for 48 hours. In that time, it designed, synthesised, and tested a new coolant for high-performance data center chips. It searched 2.7 million molecular candidates, controlled an automated lab, and completed four full cycles of design, synthesis, and measurement.
Qwen4-Max also worked with PharmMaster, a platform from Beijing-based AI drug discovery startup DP Technology. The target was a compound aimed at an inflammatory response linked to Alzheimer’s disease. Over several rounds of literature review, design, and computation, the model optimised 24 candidate molecules. According to Alibaba, the final candidate binds to the target protein about 3.6 times better than the known starting compound from a patent. DP Technology’s experts consider the result promising enough to pursue further lab tests.
Qwen4-Max aims to build playable games from a single prompt
Alibaba also shows Qwen4-Max working with established game engines like Unity, Unreal Engine, and Godot. From a single natural-language description, the model is supposed to handle the entire development process. That covers scene setup, game logic, and matching external assets, along with materials, UI, and visual effects, until the game is playable.
During development, the model runs the game itself and uses visual and interactive feedback to tweak camera work, character animations, hit reactions, and lighting. Alibaba’s sample project is “Meow of Magic,” an Unreal Engine adventure about a cat wizard in which players cast spells and fight waves of enemies. According to Alibaba, Qwen4-Max also builds 3D models in Blender that can be reused for games and 3D printing.
Anthropic recently moved its coding assistant Claude Code in the opposite direction. After a wave of unofficial Halo and GTA: Vice City clones, it barred the tool from recreating existing games.
Alibaba says the model writes safer code and offers no real CBRN uplift
On the safety benchmark Last Secure Code Bench, the share of code that is both functional and secure rose from 22 to 31.8 percent, according to Alibaba. The share rated secure at all three testing levels nearly doubled, from 8 to 14.7 percent. Alibaba also tested for CBRN risks (chemical, biological, radiological, and nuclear). Internal reviewers went through the model’s full chains of thought, the company says. They found no real capacity for innovation that would give a user a significant boost in capability.
Qwen4-Max is available through QwenCloud and in QwenWork, Qoder, and QwenStudio. Alibaba plans to release smaller Qwen4 models as open source later.
In August, Alibaba released Qwen3.8-Max, its first Max model built for autonomous tasks that run for days. It also has 2.4 trillion parameters, but 95 billion are active per token. At the time, Artificial Analysis found that the model needed 64 steps per task instead of 14 and cost more than twice as much per task as Qwen3.7-Max.
In late August, Alibaba previewed the Qwen4 architecture with Qwen3.8-Flash-Next, which already used an N-gram embedding layer with 51 billion parameters. The company had released the Qwen3.8-Max weights in mid-August. For Qwen4, only the smaller models will get open weights for now.
Moonshot AI followed in October with Kimi K3.1 Pro. At 2.8 trillion parameters, it is even larger, and Moonshot says it sits just behind Claude Opus 5.5 on Terminal-Bench 4.0. Like Alibaba, DeepSeek uses a hybrid attention design with sparse attention in V4-Pro to process long contexts more cheaply. Zhipu AI took a different route and got GLM-5.3’s gains entirely from extra post-training.



