DataLearner logo
QW

Qwen3.8-Max

Reasoning modelCoding model

Qwen3.8-Max

Release date: 2026-08-03Updated: 2026-08-31Views: 4,361
Live demoGitHubHugging FaceCompare
Parameters
2.4T
Context length
1M
Multilingual
Supported
Reasoning ability
5/5

Qwen3.8-Max is a reasoning model from Alibaba, released on 2026-08-03. It accepts text, image, and video input and returns text output. The cataloged parameter count is 2.4T, with 95B active parameters per inference. The recorded context window is 1M, and the recorded maximum output is 128K. Cataloged capabilities include Reasoning model, Multilingual, Coding model, and Function calling. The checkpoint is listed under the Qwen3 8 Max License, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

Qwen3.8-Max

Model basics

Reasoning traces
Supported
Thinking modes
Thinking Level · Extra-High (Default)Thinking Level · LowThinking Level · High
Context length
1M tokens
Max output length
128K tokens
Model type
Reasoning model
Modality (in / out)
Text, Image, Video → Text
Release date
2026-08-03
Model file size
No data
MoE architecture
Yes
Total params / Active params
2.4T / 95B
Knowledge cutoff
No data
Qwen3.8-Max

Open source & experience

Code license
Weights license
- Commercial use permitted, with conditions
GitHub repo
N/A
Hugging Face
huggingface.co/Qwen/Qwen3.8-2.4T-A95B
Live demo
qwen.ai
Qwen3.8-Max

Official resources

Paper
Paper
DataLearnerAI blog
N/A
Qwen3.8-Max

API details

API speed
3/5
💡Default unit: $/1M tokens. If vendors use other units, follow their published pricing.
Learn about pricing modes
Standard
TypeConditionInputOutput
Text-¥12.00/ 1M tokens¥36.00/ 1M tokens
international
TypeConditionInputOutput
Text-$2.00/ 1M tokens$6.00/ 1M tokens
Cache PricingPrompt Cache
TypeTTLWriteRead
Text-—¥1.50/ 1M tokens

“—” means the modality is not billed in that direction, or the vendor has not published a price for it.

Qwen3.8-Max

Benchmark Results

Qwen3.8-Max currently shows benchmark results led by Context Arena (3 / 126, score 96.92), LongBench v2 (1 / 14, score 66.30), HLE (18 / 197, score 56.20). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.

Thinking
Tool usage

General Knowledge

2 evaluations
Benchmark / mode
Score
Rank/total
HLE
Extra-High
43.60
57 / 197
HLE
Extra-HighTools
56.20
18 / 197

General Evaluation

1 evaluations
Benchmark / mode
Score
Rank/total
GPQA Diamond
Extra-High
92.60
26 / 274

Writing and Creative Capabilities

1 evaluations
Benchmark / mode
Score
Rank/total
Creative Writing
Standard Mode
1840.90
13 / 106

Common Sense Reasoning

1 evaluations
Benchmark / mode
Score
Rank/total
SimpleBench
Extra-High
62.50
24 / 93

Coding and Software Engineer

5 evaluations
Benchmark / mode
Score
Rank/total
FrontierSWE
Extra-HighTools
73.50
4 / 4
SWE-Bench Pro - Public
Extra-HighTools
67.70
6 / 62
DeepSWE
Extra-HighTools
56.60
25 / 38
NL2Repo-Bench
Extra-HighTools
55.90
11 / 16
MLS Bench
Extra-HighTools
41
3 / 5

Text Embedding

3 evaluations
Benchmark / mode
Score
Rank/total
Context Arena
Low
92.79
9 / 126
Context Arena
Medium
95.15
5 / 126
Context Arena
Extra-High
96.92
3 / 126

AI Agent - Tool Usage

2 evaluations
Benchmark / mode
Score
Rank/total
Terminal-Bench 2.1
Extra-HighTools
86.60
13 / 53
Toolathlon-Verified
Extra-HighTools
72.50
10 / 11

Long Context

1 evaluations
Benchmark / mode
Score
Rank/total
LongBench v2
Extra-High
66.30
1 / 14

Agent Level Benchmark

1 evaluations
Benchmark / mode
Score
Rank/total
Agents' Last Exam
Extra-HighTools
27
11 / 19

Productivity Knowledge

1 evaluations
Benchmark / mode
Score
Rank/total
AutomationBench
Extra-HighTools
27.30
15 / 17

Math and Reasoning

2 evaluations
Benchmark / mode
Score
Rank/total
FrontierMath v2
Extra-High
74.74
11 / 58
FrontierMath Tier 4 v2
Extra-High
46.34
13 / 42
View benchmark analysisCompare with other models

Compare with other models

Qwen3.8-Max

Publisher

Qwen3.8-Max

Model Overview

Qwen3.8-Max is a reasoning model from Alibaba, released on 2026-08-03.

It accepts text, image, and video input and produces text output. Its cataloged capabilities include Reasoning model, Multilingual, Coding model, Function calling, and Tool use. The cataloged parameter count is 2.4T, with 95B active parameters per inference. The recorded context window is 1M, and the recorded maximum output is 128K.

The checkpoint is listed under the Qwen3 8 Max License, with the license link included in the model references. The page records 5 API pricing rules from the listed provider; current provider pricing and conditions should be checked before deployment. The evaluation section contains 15 cataloged benchmark results with their recorded modes and scores. The page links 3 release, model-card, repository, or provider references for checking the underlying claims. Specifications, availability, and prices can change; undisclosed values are intentionally left unstated.

Qwen3.8-Max

FAQ

What is Qwen3.8-Max?

Qwen3.8-Max is a reasoning model from Alibaba, released on 2026-08-03. It accepts text, image, and video input and returns text output. The cataloged parameter count is 2.4T, with 95B active parameters per inference. The recorded context window is 1M, and the recorded maximum output is 128K. Cataloged capabilities include Reasoning model, Multilingual, Coding model, and Function calling. The checkpoint is listed under the Qwen3 8 Max License, with the license link included in the model references. Use the linked references to confirm current access, licensing, and provider-specific limits.

What input and output modalities does Qwen3.8-Max support?

The current model record lists text, image, and video as input and text as output.

What are the main recorded specifications for Qwen3.8-Max?

The cataloged parameter count is 2.4T, with 95B active parameters per inference. The recorded context window is 1M, and the recorded maximum output is 128K. Fields without a source-backed value remain undisclosed.

Does Qwen3.8-Max have API pricing?

The page records 5 API pricing rules from the listed provider; current provider pricing and conditions should be checked before deployment.

Are benchmark results available for Qwen3.8-Max?

The evaluation section contains 15 cataloged benchmark results with their recorded modes and scores. Compare only results that use the same benchmark version and evaluation mode.

Is Qwen3.8-Max open source?

The checkpoint is listed under the Qwen3 8 Max License, with the license link included in the model references. Review the linked license text before commercial or derivative use.

DataLearner on WeChat

Follow DataLearner on WeChat for AI model updates and research notes.

DataLearner WeChat QR code