Alibaba opens a two-week preview of Qwen4-Max with a 1M-token context, but no benchmarks or weights
Alibaba Cloud has opened a time-limited preview of Qwen4-Max, its next flagship model, through the Token Plan only. It offers a 1M-token context and up to 393,216 output tokens, but the documentation lists no pricing, benchmarks or downloadable weights, and the model retires on 25 November.
Alibaba Cloud has begun a preview of Qwen4-Max, the next flagship in its Qwen family, ahead of a full Qwen 4 launch that Alibaba has so far described only as coming. Third-party trackers date the preview to 11 November.
According to Alibaba Cloud's Model Studio documentation, the model ID is qwen4-max-preview and it supports a context window of one million tokens. Maximum output is 393,216 tokens, shared between the model's reasoning and its final answer, so developers who size limits only for answers may see truncated responses.
Access is restricted to Alibaba's Token Plan, in personal or team form, through a dedicated base URL. Pay-as-you-go access is not offered. Thinking is switched on by default, with a reasoning-effort setting of low, medium or xhigh, the last being the default.
The preview is deliberately short: the documentation says it ends at 24:00 on 25 November 2026 (UTC+8), after which the model is retired, and that no changes will be made during the period. The documentation does not mention pricing or benchmark results, and independent write-ups report that no model card, parameter count, licence or downloadable weights have been published.
Why it matters
Alibaba has made open weights a hallmark of Qwen, so a closed, metered preview of the top model is notable. It suggests the flagship tier is being run more like a commercial service while smaller models carry the open-source reputation.
With no published evaluations, claims about its quality cannot yet be checked, and the one-million-token window is the only hard capability figure available.
Indicops view
Treat this as a model to evaluate, not to adopt. Indian teams can test it on non-sensitive workloads, but should note the endpoint is hosted on Alibaba Cloud infrastructure and the model disappears within weeks. Keep production designs model-agnostic and apply data-residency and vendor-risk reviews before any commitment.
More in Models
Back to the newsroomGoogle launches Gemini 4 Hydrogen, a $1.50-per-million-token model for coding and agents
Google has released Gemini 4 Hydrogen, a faster and cheaper tier of its Gemini 4 family aimed at everyday coding and agentic work. It costs $1.50 per million input tokens and $7.50 per million output tokens across a 1M-token window, and is rolling out through the Gemini API, AI Studio and consumer products.
Anthropic releases Claude Fable 5.2 and Mythos 5.2, 30 per cent faster and 20 per cent cheaper
Anthropic has released Claude Fable 5.2 for general use and Mythos 5.2 for vetted cyber and life-sciences organisations. The company says output is up to 30 per cent faster than the 5.1 generation, at $8 per million input tokens and $40 per million output tokens, with lower rates of safeguard blocks.
Meta launches Muse Ember, a low-cost agent model priced at $0.10 per million input tokens
Meta has released Muse Ember, the smallest model in its Muse family, aimed at high-volume agent work such as tool calls, extraction and routing. It offers a 1M-token context and 512K output tokens, priced from $0.10 per million input tokens, and is available through Meta's API and Google and Oracle clouds.