Alibaba Drops Qwen4-Max: 1M Context, Half Price, and Zero Open Weights
Alibaba casually crashed the artificial intelligence party by dropping a massive new model that eats fewer digital cookies while chatting smarter. Budget spreadsheets everywhere are quietly weeping tears of joy.
Alibaba officially unveiled Qwen4-Max, bringing a hefty one-million token context window and native image processing directly to the corporate playground. Internal benchmarks show the model devours up to fifty percent fewer tokens compared to its predecessor Qwen3.8-Max, depending heavily on the specific workload. API pricing sits at $1.50 per million input tokens and $4.50 for output tokens, while cached inputs drop to a microscopic seven and a half cents.
Task efficiency soared during internal tests, where the average cost per task on Work360 plummeted from $4.43 down to $1.67. On the coding leaderboard Code Arena WebDev, the system debuted at sixth place with 1750 points, comfortably outpacing rival systems like Kimi K3.1 Pro and GLM-5.3. Western heavyweights such as GPT-6.1 Sol and GPT-6 Astra hover nearby, while elite champions like Gemini 4 Carbon and Opus 5.5 remain parked slightly higher.
Real-world agent evaluations placed the flagship model at fourteenth place in Agent Arena, marking the highest score achieved by any regional competitor in that category. While the gargantuan Max variant keeps its inner workings strictly locked away behind corporate walls, smaller sibling models with open weights are scheduled for release soon. According to hardware optimization notes from Unsloth, the upcoming Qwen4-27B variant will comfortably fit onto a standard consumer machine equipped with sixteen gigabytes of RAM.
The massive flagship infrastructure is already live across multiple developer gateways including OpenCode Go, Qwen Studio, Qoder, and the dedicated cloud interface.
Another proprietary giant locks the source code while democratizing the lighter versions just enough to keep the open-source community perpetually hungry and grateful.
Source: Qwen Blog
Comments
This is where the magic happens: AI reads your discussion and rewrites the article based on the most interesting comments. Each strong comment adds points to the meter below. Once the meter is full, the article updates live — no page reload needed.