Model

Qwen3.8-Plus

CLOSED

Alibaba · Qwen3.8 family · released Sep 9, 2026

Alibaba's September 9, 2026 mid-tier API model, slotted between Qwen3.8-Flash-Next and the flagship 0902 Max refresh — the hosted counterpart to the open 27B weights. Priced at $0.80/$2.40 per Mtok, roughly 2.5x below the Max, with the same 1M context. Launch table is entirely vendor-run (Qwen's own harness); it leads the open 27B across the board but trails the Max by 3-5 points on the shared rows, matching the historical Plus-vs-Max gap. Video-MME Pro row is the first filing on that benchmark revision for this catalog — Qwen is the only vendor citing it so far.

ReasoningCodingVisionFunction callingTool useAgentic

2392.5

Elo · rank #51

Parameters

Undisclosed

Active params

Undisclosed

Context

1M tokens

Architecture

API-only mid tier of the Qwen3.8 line (hybrid linear/full-attention stack shared with the open 27B); undisclosed parameter count, 1M-token context

License

Proprietary

Languages

—

API price (in/out)

$0.8 / $2.4

Modalities

text · vision

Benchmark results

Bar shows position within the tracked field; marker = field best

DeepSWECoding58.9%#13

best: Muse Spark 1.3 · 75.4%

GDPval-AAAgents1492#28

best: Claude Fable 5 · 1932

GPQA DiamondReasoning89.4%#41

best: GPT-6 Astra · 96.0%

Humanity's Last ExamReasoning41.2%#32

best: Claude Opus 5 · 64.7%

JobBenchAgents55.7%#6

best: Claude Opus 5 · 65.7%

LiveCodeBenchCoding78.9%#60

best: DeepSeek-V4.1 (Think Max) · 94.2%

MCP AtlasAgents76.8%#5

best: Kimi K3.1 · 85.6%

MMMU-ProVision80.6%#13

best: Claude Opus 4.7 · 85.5%

OSWorld 2.0Agents52.4%#8

best: Claude Fable 5.1 · 77.9%

SWE-bench ProCoding60.2%#21

best: Claude Fable 5.1 · 81.2%

Terminal-Bench 2.0Coding74.8%#23

best: Gemini 3.8 Flash · 89.4%

Video-MME ProVision84.3%#2

best: Seed 2.2 Pro · 90.4%

Run it locally

Closed weights — available via API only. No local deployment.

Input / M tok

$0.8

Output / M tok

$2.4

Qwen3.8 family

Elo progression across releases

Qwen3.8 MaxAug 20262833.6 Qwen3.8-27BAug 20262461.0 Qwen3.8-Flash-NextAug 20262613.7 Qwen3.8-Max-0902Sep 20263032.6 Qwen3.8-PlusSep 20262392.5

Resources

DOCShttps://qwen.ai/blog?id=qwen3.8-plus APIAvailable via Alibaba API

API price $0.8/$2.4 · each benchmark row carries its own source badge (see methodology)