EN한
← wiki

$ cat wiki/models/qwen4-max.md

Qwen4-Max

modelupdated 2026-11-19created 2026-11-19

Compared with

Spec

AttributeValue
DeveloperAlibaba / Qwen AI Lab
ReleasedAvailable by 2026-11-19 (reference checked)
Announced2026-11-16 (indexed official announcement)
Context window1,000,000 tokens
PricingSingapore International: $1.5/M input · $4.5/M output, thinking and non-thinking
Licenseunknown
AvailabilityAlibaba Cloud Model Studio API; model ID qwen4-max
The quoted rates are for the Singapore International deployment and input requests up to 1M tokens; output includes chain-of-thought and response. Other regional tables differ. The reference lists no fine-tuning or model deployment support. (source)

Release Date

The official announcement's indexed title/date identifies 2026-11-16. Its direct page returned an empty application shell during this review; the substantive specifications here come from the separately fetched model reference on 2026-11-19. (source) (source)

Benchmarks

No comparative score is adopted from the reviewed reference. The launch article's indexed performance claims are not used to rank this model. (source) (source)

Use Cases

The reference documents text, image and video inputs with text output, web search, parallel tool calls and JSON Schema output. It supports OpenAI Chat Completions, Responses, Anthropic Messages and DashScope protocols. Reasoning effort is low, high or xhigh, defaulting to high. (source)

Compared To

Like Gemini 4 Neon, Qwen4-Max targets multimodal agent work. Protocol support establishes an integration route, not equal task performance or identical reasoning semantics across providers. (source) (source)

Sources

Referenced by

Sources