Thinking-enabled DeepSeek-V3.1-Terminus with improved language consistency, upgraded Code/Search Agents, and stronger stability and reliability versus V3.1. FP8.
Context Window
128.0K
Max Output
65.5K
Avg output tokens (7d)
548 tokens
Input Price (Auto)
$0.25/1M
Output Price (Auto)
$0.70/1M
Cache Read (Auto)
$0.13/1M
Capabilities
Benchmarks
Benchmarks
Performance metrics and benchmarks
Sourced from Artificial Analysis.
Intelligence Index
15.4
Coding Index
43.5
Agentic work
AutomationBench-AA
Workflow automation with guardrail penalties
2.7%
Better than 26% of models compared
AA-Briefcase
Agentic knowledge work (Elo)
464 Elo
Better than 24% of models compared
GDPval-AA v2
Economically valuable tasks (Elo)
826 Elo
Better than 38% of models compared
Document reasoning
GDP.pdf
Professional PDF reasoning: all-pass rate
5.4%
Better than 31% of models compared
AA-LCR v1.1
Long context reasoning with updated grading
69.3%
Better than 65% of models compared
Reasoning
HLE
Humanity's Last Exam
16.4%
Better than 67% of models compared
IFBench
Instruction-following benchmark
57.0%
Better than 68% of models compared
Coding
Terminal-Bench v4.0
Practical coding and terminal tasks
0.0%
Better than 17% of models compared
SciCode
Python programming for scientific computing
38.0%
Better than 21% of models compared
LiveCodeBench
Contamination-free coding benchmark
79.8%
Better than 92% of models compared
Math
AIME 2025
American Invitational Mathematics Examination 2025
89.7%
Better than 90% of models compared
Knowledge
MMLU-Pro
Professional and academic subject knowledge
85.1%
Better than 91% of models compared
Legacy benchmarks
GPQA Diamond (legacy)
Graduate-level scientific reasoning
79.2%
Better than 66% of models compared
Terminal-Bench Hard (legacy)
Agentic coding and terminal use
30.3%
Better than 72% of models compared
T²-Bench Telecom (legacy)
Conversational AI agents in dual-control scenarios
37.1%
Better than 45% of models compared
AA-LCR (unversioned / legacy)
Long context reasoning evaluation
69.3%
Better than 65% of models compared
Last updated Sep 13, 2026
Artificial AnalysisProviders
Choose explicit providers for this model. Auto routing remains available as the default option.
Loading provider options…
Related text models
Compare DeepSeek V3.1 Terminus (Thinking) with similar models from the same provider or model family.
DeepSeek V3.1 Terminus
deepseek-ai/DeepSeek-V3.1-TerminusDeepSeek-V3.1-Terminus. The latest update builds on V3.1's strengths while addressing key user feedback. Language consistency improvements (fewer CN/EN mix-ups, no random chars), stronger Code Agent & Search Agent performance, and more stable, reliable outputs across benchmarks. FP8.
DeepSeek V3.1
deepseek-ai/DeepSeek-V3.1DeepSeek-V3.1 is a hybrid model that supports both thinking mode and non-thinking mode. It does better at tool calling and agent tasks, and has higher thinking efficiency than its predecessor. This is the non-thinking version. Quantized at FP8.
DeepSeek R1 0528
deepseek-ai/DeepSeek-R1-0528The new (May 28th) Deepseek R1 model.
DeepSeek V3.1 Thinking
deepseek-ai/DeepSeek-V3.1:thinkingThinking enabled version of Deepseek V3.1. DeepSeek-V3.1 is a hybrid model that supports both thinking mode and non-thinking mode. It does better at tool calling and agent tasks, and has higher thinking efficiency than its predecessor. Quantized at FP8.
DeepSeek V3.2 Exp
deepseek-ai/deepseek-v3.2-expDeepSeek V3.2 Exp is an experimental V3-series model with improved long-context efficiency compared with earlier V3 releases. FP8.
DeepSeek V3.2 Exp Thinking
deepseek-ai/deepseek-v3.2-exp-thinkingDeepSeek V3.2 Exp Thinking is an experimental V3-series reasoning model with improved long-context efficiency compared with earlier V3 releases. FP8.