DataLearner logo
DE

DeepSeek-V3.1

Chat modelDeepSeek VDeepSeek V3.1

DeepSeek-V3.1

Release date: 2025-08-20Updated: 2026-07-17Views: 2,783
Live demoGitHubHugging FaceCompare
Parameters
671B
Context length
128K
Multilingual
Supported
Reasoning ability
4/5

DeepSeek-V3.1 is a chat model from DeepSeek-AI, released on 2025-08-20. It accepts text input and returns text output. The cataloged parameter count is 671B, with 37B active parameters per inference. The recorded context window is 128K, and the recorded maximum output is 8K. Cataloged capabilities include Reasoning model and Multilingual. The checkpoint is listed under the MIT 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

DeepSeek-V3.1

Model basics

Reasoning traces
Supported
Thinking modes
Standard Mode (Default)Thinking Mode
Context length
128K tokens
Max output length
8K tokens
Model type
Chat model
Modality (in / out)
Text → Text
Release date
2025-08-20
Model file size
1340GB
MoE architecture
Yes
Total params / Active params
671B / 37B
Knowledge cutoff
No data
DeepSeek-V3.1

Open source & experience

Code license
MIT License
Weights license
MIT License- Commercial use permitted
GitHub repo
N/A
Hugging Face
huggingface.co/deepseek-ai/DeepSeek-V3.1-Base
Live demo
chat.deepseek.com
DeepSeek-V3.1

Official resources

Paper
Paper
DataLearnerAI blog
DataLearnerAI blog
DeepSeek-V3.1

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-$0.560/ 1M tokens$1.68/ 1M tokens
Batch
TypeConditionInputOutput
Text-$0.280/ 1M tokens$0.840/ 1M tokens
Cache PricingPrompt Cache
TypeTTLWriteRead
Text-$0.560/ 1M tokens$0.280/ 1M tokens
Text-$0.560/ 1M tokens
Cache = write
—
Text-—$0.280/ 1M tokens
Cache = hit

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

DeepSeek-V3.1

Benchmark Results

DeepSeek-V3.1 currently shows benchmark results led by MMLU (1 / 124, score 93.40), SimpleQA (4 / 47, score 93.40), AIME 2024 (7 / 62, score 93.10). 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

5 evaluations
Benchmark / mode
Score
Rank/total
MMLU
Standard Mode
91.80
4 / 124
MMLU
Thinking Mode
93.40
1 / 124
MMLU Pro
Standard Mode
83.70
44 / 134
MMLU Pro
Thinking Mode
85
27 / 134
HLE
Thinking Mode
15.90
156 / 197

General Evaluation

2 evaluations
Benchmark / mode
Score
Rank/total
GPQA Diamond
Standard Mode
74.90
175 / 274
GPQA Diamond
Thinking Mode
80.10
146 / 274

Common Sense

1 evaluations
Benchmark / mode
Score
Rank/total
SimpleQA
Thinking Mode
93.40
4 / 47

Coding and Software Engineer

3 evaluations
Benchmark / mode
Score
Rank/total
LiveCodeBench
Standard Mode
56.40
84 / 128
LiveCodeBench
Thinking Mode
74.80
44 / 128
SWE-bench Verified
Standard Mode
66
76 / 116

Math and Reasoning

4 evaluations
Benchmark / mode
Score
Rank/total
AIME 2024
Standard Mode
66.30
40 / 62
AIME 2024
Thinking Mode
93.10
7 / 62
AIME2025
Standard Mode
49.80
88 / 107
AIME2025
Thinking Mode
88.40
42 / 107

Writing and Creative Capabilities

1 evaluations
Benchmark / mode
Score
Rank/total
Creative Writing
Standard Mode
1433.20
57 / 106

AI Agent - Tool Usage

1 evaluations
Benchmark / mode
Score
Rank/total
Terminal-Bench
Standard ModeTools
31.30
19 / 35

Common Sense Reasoning

1 evaluations
Benchmark / mode
Score
Rank/total
SimpleBench
Standard Mode
40
68 / 94
View benchmark analysisCompare with other models
DeepSeek-V3.1

Publisher

DeepSeek-V3.1

Model Overview

DeepSeek-V3.1 is a chat model from DeepSeek-AI, released on 2025-08-20.

It accepts text input and produces text output. Its cataloged capabilities include Reasoning model and Multilingual. The cataloged parameter count is 671B, with 37B active parameters per inference. The recorded context window is 128K, and the recorded maximum output is 8K.

The checkpoint is listed under the MIT license, with the license link included in the model references. The page records 8 API pricing rules from Fireworks AI; current provider pricing and conditions should be checked before deployment. The evaluation section contains 17 cataloged benchmark results with their recorded modes and scores. The page links 4 release, model-card, repository, or provider references for checking the underlying claims. Specifications, availability, and prices can change; undisclosed values are intentionally left unstated.

DeepSeek-V3.1

FAQ

What is DeepSeek-V3.1?

DeepSeek-V3.1 is a chat model from DeepSeek-AI, released on 2025-08-20. It accepts text input and returns text output. The cataloged parameter count is 671B, with 37B active parameters per inference. The recorded context window is 128K, and the recorded maximum output is 8K. Cataloged capabilities include Reasoning model and Multilingual. The checkpoint is listed under the MIT 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 DeepSeek-V3.1 support?

The current model record lists text as input and text as output.

What are the main recorded specifications for DeepSeek-V3.1?

The cataloged parameter count is 671B, with 37B active parameters per inference. The recorded context window is 128K, and the recorded maximum output is 8K. Fields without a source-backed value remain undisclosed.

Does DeepSeek-V3.1 have API pricing?

The page records 8 API pricing rules from Fireworks AI; current provider pricing and conditions should be checked before deployment.

Are benchmark results available for DeepSeek-V3.1?

The evaluation section contains 17 cataloged benchmark results with their recorded modes and scores. Compare only results that use the same benchmark version and evaluation mode.

Is DeepSeek-V3.1 open source?

The checkpoint is listed under the MIT 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