Qwen3.8 Max

Name: Qwen3.8 Max
Author: Alibaba

Multimodal

Alibaba

Qwen3.8 Max is Qwen's 2.4T-parameter multimodal flagship released in August 2026. It is a state-of-the-art performer on research replication (PaperBench 0.93) and long-context retrieval (MRCR v2 8-needle 0.93). It is served by DeepInfra, Fireworks, Novita and Together.

Key Specifications

Parameters

2.4T

Context

1.0M

Release Date

August 2, 2026

Average Score

71.0%

API Documentation Research Paper Results Blog

Timeline

Key dates in the model's history

Announcement

August 2, 2026

Last Update

August 27, 2026

Today

September 10, 2026

Technical Specifications

Parameters

2.4T

Training Tokens

Knowledge Cutoff

Family

Capabilities

MultimodalZeroEval

Pricing & Availability

Input (per 1M tokens)

$2.50

Output (per 1M tokens)

$6.25

Max Input Tokens

1.0M

Max Output Tokens

131.1K

Supported Features

Function CallingStructured OutputCode ExecutionWeb SearchBatch InferenceFine-tuning

Benchmark Results

Model performance metrics across various tests and benchmarks

Reasoning

Logical reasoning and analysis

GPQA

• Self-reported

92.6%

Other Tests

Specialized benchmarks

PaperBench

BasicAgent Code-Dev, average of 3 runs • Self-reported

93.0%

MRCR v2 (8-needle)

256K context; 8 needles • Self-reported

93.0%

Agents' Last Exam

Score • Self-reported

52.4%

AndroidBench

• Self-reported

75.1%

AndroidWorld

• Self-reported

85.3%

AutomationBench

Pass@1 • Self-reported

27.3%

CoWorkBench

• Self-reported

74.8%

DeepSWE 1.1

Best result across Claude Code and mini-SWE-agent • Self-reported

56.6%

ERQA

• Self-reported

77.8%

FrontierSWE

Claude Code, MEAN@5 • Self-reported

73.5%

HealthBench

• Self-reported

60.2%

Humanity's Last Exam

• Self-reported

43.6%

Humanity's Last Exam (with tools, text-only)

With tools • Self-reported

56.2%

IFBench

• Self-reported

82.8%

Job Bench

• Self-reported

53.4%

LongBench v2

• Self-reported

66.3%

LVBench

• Self-reported

81.8%

MLS-Bench Lite

Claude Code, 5-hour timeout • Self-reported

41.0%

MMMU-Pro

Qwen in-house evaluation • Self-reported

82.3%

MobileWorld

• Self-reported

77.8%

NL2Repo

Claude Code harness • Self-reported

55.9%

OneMillion Bench

Expert score • Self-reported

52.5%

OSWorld-Verified

• Self-reported

86.1%

PerceptionBench

Qwen in-house evaluation • Self-reported

63.5%

PLawBench

• Self-reported

73.2%

PRBench-Finance

Finance subset • Self-reported

58.3%

PRBench-Legal

Legal subset • Self-reported

57.6%

QwenQoderBench

• Self-reported

58.4%

QwenReactBench

BT/Elo rating • Self-reported

86.2%

QwenSVG

BT/Elo rating • Self-reported

85.7%

QwenSWEBench

• Self-reported

80.7%

RealWorldQA

• Self-reported

88.0%

ScreenSpot Pro

Qwen in-house evaluation • Self-reported

84.5%

SkillsBench

SkillsBench v1.1, OpenCode, average of 3 runs • Self-reported

70.2%

SWE-Bench Pro

Claude Code, 256K context • Self-reported

67.7%

Terminal-Bench 2.1

Claude Code, avg@10, 5-hour timeout • Self-reported

86.6%

Toolathlon

Verified split; Pass@1 • Self-reported

72.5%

VideoMME w sub.

Subtitles enabled • Self-reported

90.4%

Vision2Web

Claude Code, GPT-5.4 judge • Self-reported

69.0%

WideSearch

• Self-reported

81.9%

Workspace Bench

• Self-reported

67.7%

License & Metadata

License

proprietary

Announcement Date

August 2, 2026

Last Updated

August 27, 2026

Compare Qwen3.8 Max

All comparisons

vs Qwen3.8 Flash vs Qwen3.8 Flash Next vs Qwen3.8-27B vs Muse Spark 1.2 vs Muse Spark 1.1 vs Solar Pro 4 vs Muse Spark 1.3 vs Ling 3.0 Flash

Similar Models

All Models

Qwen3.5-397B-A17B

Alibaba

MM397.0B

Best score:0.9 (GPQA)

Released:Feb 2026

Qwen3.8 Flash

Alibaba

MM125.0B

Best score:0.9 (GPQA)

Released:Aug 2026

Price:$0.15/1M tokens

Qwen3.8 Flash Next

Alibaba

MM180.0B

Best score:0.9 (GPQA)

Released:Aug 2026

Price:$0.16/1M tokens

Qwen3.6 Plus

Alibaba

Best score:0.9 (GPQA)

Released:Mar 2026

Price:$0.50/1M tokens

Qwen3.8-27B

Alibaba

MM27.8B

Best score:0.9 (GPQA)

Released:Aug 2026

Qwen3.6-27B

Alibaba

MM27.8B

Best score:0.9 (GPQA)

Released:Apr 2026

Price:$0.60/1M tokens

Qwen3.7-Plus

Alibaba

Best score:0.9 (GPQA)

Released:May 2026

Price:$0.32/1M tokens

Qwen3.6-35B-A3B

Alibaba

MM35.0B

Best score:0.9 (GPQA)

Released:Apr 2026

Recommendations are based on similarity of characteristics: developer organization, multimodality, parameter size, and benchmark performance. Choose a model to compare or go to the full catalog to browse all available AI models.