█

AI/TLDR

Buy me a coffee
Buy me a coffee
PROMOTE YOUR WEBSITE IN JUST $20

Model comparison

DeepSeek-V3.2-Exp vs DeepSeek-V3.1-Terminus

DeepSeekDeepSeek-V3.2-Exp
Input
$0.28 / 1M tokens (cache miss)
Output
$0.42 / 1M tokens
Context
128K
Released
2025-09-29
Full model page →DeepSeekDeepSeek-V3.1-Terminus
Input
$0.27 / 1M tokens
Output
$0.95 / 1M tokens
Context
128K
Released
2025-09-22
Full model page →
Short answer

DeepSeek-V3.2-Exp and DeepSeek-V3.1-Terminus split the 9 benchmarks both report evenly; DeepSeek-V3.1-Terminus is cheaper on input tokens ($0.3 vs $0.3 per 1M, 1×).

Specs and pricing side by side

SpecDeepSeek-V3.2-ExpDeepSeek-V3.1-Terminus
MakerDeepSeekDeepSeeksame
Released2025-09-292025-09-22
Context window128K128Ksame
Max output64K32K
Input price$0.28 / 1M tokens (cache miss)$0.27 / 1M tokens
Output price$0.42 / 1M tokens$0.95 / 1M tokens
Parameters685B total (~37B active, MoE)685B total, 37B active (MoE)
LicenseMITMITsame
WeightsOpen weightsOpen weightssame
ModalitiesTextTextsame

Not published by either maker: knowledge cutoff.

Benchmarks both models report

BenchmarkDeepSeek-V3.2-ExpDeepSeek-V3.1-Terminus
MMLU-Pro85%85%
GPQA-Diamond79.9%80.7%
Humanity's Last Exam19.8%21.7%
LiveCodeBench74.1%74.9%
Aider-Polyglot74.5%76.1%
BrowseComp40.1%38.5%
SimpleQA97.1%96.8%
SWE-bench Multilingual57.9%57.8%
Terminal-bench37.7%36.7%

Scores are as published by each maker or its cited source; harness, reasoning effort and date can differ between rows, so treat small gaps as ties. Sources are listed on each model page.

Which one should you pick?

Pick DeepSeek-V3.2-Exp if…

  • You care about BrowseComp, SimpleQA, SWE-bench Multilingual (it scores higher on 4 shared benchmarks).
  • Your use case is: long-context document, codebase, and transcript analysis where attention cost dominates.

Pick DeepSeek-V3.1-Terminus if…

  • You care about GPQA-Diamond, Humanity's Last Exam, LiveCodeBench (it scores higher on 4 shared benchmarks).
  • Cost matters: input tokens cost 1× less ($0.3 vs $0.3 per 1M).
  • Your use case is: agentic coding assistants and autonomous software-engineering loops (SWE-bench-style tasks).

FAQ

Which is better, DeepSeek-V3.2-Exp or DeepSeek-V3.1-Terminus?

DeepSeek-V3.2-Exp and DeepSeek-V3.1-Terminus split the 9 benchmarks both report evenly; DeepSeek-V3.1-Terminus is cheaper on input tokens ($0.3 vs $0.3 per 1M, 1×). The better pick depends on whether you value benchmark scores, price per token, context length or openness for your workload.

How much cheaper is DeepSeek-V3.1-Terminus?

DeepSeek-V3.2-Exp: $0.28 / 1M tokens (cache miss) input, $0.42 / 1M tokens output. DeepSeek-V3.1-Terminus: $0.27 / 1M tokens input, $0.95 / 1M tokens output (list API prices as published by each maker).

What are the context windows of DeepSeek-V3.2-Exp and DeepSeek-V3.1-Terminus?

DeepSeek-V3.2-Exp offers 128K with up to 64K output; DeepSeek-V3.1-Terminus offers 128K with up to 32K output.

Is DeepSeek-V3.2-Exp newer than DeepSeek-V3.1-Terminus?

DeepSeek-V3.2-Exp was released 2025-09-29; DeepSeek-V3.1-Terminus was released 2025-09-22.

Go deeper