Alibaba has launched Qwen4-Max, a large mixture-of-experts model featuring a 2.4 trillion parameter backbone with 71 billion active parameters per token. This new model boasts a 1 million token context window and native multimodal capabilities, accepting text, images, and video. While the flagship Qwen4-Max is proprietary and available via API, Alibaba has also released smaller open-weight Qwen4 models. The company reports significant cost reductions for API usage, with average task costs dropping by approximately 62% due to fewer tokens used and lower API prices. AI
IMPACT Sets a new benchmark for context window size and multimodal capabilities, potentially driving down costs for agentic workloads.
RANK_REASON Frontier-lab model release with system card and performance benchmarks.
Read on Mastodon — mastodon.social →
- Alibaba Group
- qwen4-max
- Claude Opus 5.5
- GPT-6.1 Sol
- GPT-6 Astra
- Kimi K3.1
- Qoder
- Qwen3.8-Max
- Qwen4
- QwenCloud
- Qwen Studio
- QwenWork
- Work360
- Claude Fable 5.1
- Hugging Face
AI-generated summary · Google Gemini · from 10 sources. How we write summaries →