Model

Qwen3.5-Omni-Plus

CLOSED

Alibaba · Qwen3.5 family · released Mar 30, 2026

Flagship tier of the Qwen3.5-Omni family; claims SOTA on 215 audio/audio-visual understanding, reasoning, and interaction subtasks/benchmarks (3 audio-visual, 5 audio, 8 ASR, 156 language-specific S2TT, 43 language-specific ASR benchmarks, per the launch blog). Supports controllable, structured audio/audio-visual captioning down to screenplay-level detail.

ReasoningCodingVisionFunction callingTool useAgentic

1834.0

Elo · rank #77

Parameters

Undisclosed

Active params

Undisclosed

Context

256K tokens

Architecture

Thinker-Talker omnimodal architecture; both Thinker and Talker adopt Hybrid-Attention MoE; RVQ speech tokenizer plus ARIA text-speech interleaving in the Talker; exact parameter/expert configuration undisclosed

License

Proprietary

Languages

—

API price (in/out)

$1.4 / $8.3

Modalities

text · vision · audio · video

Benchmark results

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

AI2DVision91.2%#19

best: Molmo 72B · 96.3%

C-EvalKnowledge92.0%#5

best: Qwen3.6-Plus · 93.3%

CharXivVision72.5%#20

best: Muse Spark 1.1 (xhigh) · 88.4%

GPQA DiamondReasoning83.9%#62

best: GPT-5.6 · 94.6%

IFBenchReasoning52.6%#34

best: MiniMax M3 · 83.0%

IFEvalReasoning89.7%#19

best: Gemma 4 26B A4B · 98.5%

LiveCodeBenchCoding65.6%#83

best: DeepSeek-V4-Pro (Think Max) · 93.5%

MathVisionVision73.0%#11

best: Seed 2.1 Pro · 92.6%

MathVistaVision86.1%#8

best: Seed 2.1 Pro · 90.7%

MMBench (English)Vision92.8%#3

best: Qwen3.5-397B-A17B · 93.7%

MMLU-ProKnowledge85.9%#23

best: Claude Fable 5 · 91.5%

MMLU-ReduxKnowledge94.2%#7

best: Qwen3.7-Max · 95.0%

MMMU-ProVision73.9%#27

best: Claude Opus 4.7 · 85.5%

MMMUVision80.1%#24

best: Claude Fable 5 · 89.3%

OCRBenchVision91#16

best: InternVL3-78B · 906

RefCOCOVision95.0%#1

best: this model · 95.0%

SuperGPQAReasoning66.4%#8

best: Qwen3.7-Max · 73.6%

Video-MMEVision81.9%#10

best: Seed 2.1 Pro · 89.2%

Run it locally

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

Input / M tok

$1.4

Output / M tok

$8.3

Qwen3.5 family

Elo progression across releases

Qwen3.5-397B-A17BFeb 20262182.0 Qwen3.5-PlusFeb 20261882.6 Qwen3.5-122B-A10BFeb 20262021.4 Qwen3.5-Omni-FlashMar 20261471.5 Qwen3.5-Omni-PlusMar 20261834.0

Resources

DOCShttps://qwen.ai/blog?id=qwen3.5-omni APIAvailable via Alibaba API

API price $1.4/$8.3 · each benchmark row carries its own source badge (see methodology)