Skip to content
CheapestInference

Qwen3.8 Max API — status, specs & subscription

Qwen3.8 Max (also written “Qwen 3.8 Max”) is Alibaba’s flagship model — a 2.4-trillion-parameter system scoring 53 on the Artificial Analysis Intelligence Index, with a 1M-token context window. At its list price it sits on the price-vs-intelligence Pareto frontier — the full analysis, the head-to-head with Kimi K3, and the chart are in Qwen3.8 Max: specs, benchmarks, pricing & API options.

Quick facts

Section titled “Quick facts”
ModelQwen3.8 Max
Model IDqwen3.8-max
StatusReviewing — pipeline
Pool candidateFlagship Pool (next to Kimi K3)
AA Intelligence Index53
Context window1M tokens
Parameters~2.4T (MoE, config unpublished)
ReasoningHybrid — reasons on demand
Sampling paramsStandard semantics — temperature (0–2), top_p, penalties
List price (reference)$2.00 in / $6.00 out per 1M tokens
Open weightsAnnounced, not yet released — no license, no model card, no date

Why it’s a Flagship candidate

Section titled “Why it’s a Flagship candidate”

The Flagship Pool serves the model ceiling, and Qwen3.8 Max earns the shortlist on the numbers: top-five intelligence (53, four points off Kimi K3’s 57), a 1M-token context window, and a different working profile than K3 — hybrid reasoning you can toggle per request and standard sampling controls next to K3’s fixed calibration. One subscription covering both profiles is the point.

What’s gating it is licensing, not quality: Alibaba has announced open weights but published nothing yet. Qwen’s open releases (3.5, 3.6) shipped under Apache 2.0; its Max tier has historically stayed closed. Which precedent wins here decides the outcome.

Follow the status

Section titled “Follow the status”