Deepseek V3.1 Terminus
DeepSeek develops the DeepSeek V 3.1 Terminus, a chat model exceling at tasks such as code generation, function calling, and reasoning, with a notable context window of 163,840 tokens. Its Math Index score of 89.7 places it among the top 25% of models, demonstrating a particular strength in mathematical reasoning.
Input
Output
Context
164K
Max Output
33K
Parameters
684.5B
Technical Specifications
Model TypeChat
Context Window163,840 tokens
Max Output Tokens32,768 tokens
Parameters684.5B
Release DateSep 22, 2025
Training CutoffMar 1, 2025
Licensemit
Open SourceYes
Input Modalities
Text
Output Modalities
Text
Capabilities
Benchmarks
Artificial Analysis
15.2%
HLE
65.0%
AA-LCR
79.2%
GPQA
37.1%
Tau2-Bench
89.7%
AIME 2025
57.0%
IFBench
40.6%
SciCode
85.1%
MMLU Pro
89.7
Math
15.9%
Tau2 Banking
43.5
Coding
79.8%
LiveCodeBench
30.4
Intelligence
30.3%
Terminal-Bench Hard
44.9%
Terminal-Bench 2.1
Resources & Links
Estimated GPU Requirements for Deepseek V3.1 Terminus
Estimates based on INT8 quantization. Actual requirements vary by framework and configuration.
More Models by DeepSeek
Deepseek V3.1 Base131K context · 684.5B · Open weightsDeepseek Prover V1.5 SFT4K context · 6.9B · Open weightsDeepseek V3.2 Speciale164K context · 685.4B · Open weightsDeepseek Math 7B Base4K context · 7B · Open weightsDeepseek R1 Distill Qwen 32B131K context · 32.8B · Open weightsDeepseek V3 Base131K context · 684.5B · Open weights
Related Tools
Data sourced from official provider APIs and documentation
Last updated: Aug 8, 2026
Your AI stack shouldn't stand still.
Every month new models become cheaper, faster, and more capable. Inferbase ensures your application automatically benefits without changing a single API call.