Price Per TokenPrice Per Token
Deepseek
Deepseek
vs
Deepseek
Deepseek

DeepSeek V3.1 Terminus vs DeepSeek V3.2 Speciale

A detailed comparison of pricing, benchmarks, and capabilities

OpenClaw

Best LLMs for OpenClaw Vote for which model works best with OpenClaw

112 out of our 301 tracked models have had a price change in February.

Get our weekly newsletter on pricing changes, new releases, and tools.

Key Takeaways

DeepSeek V3.1 Terminus wins:

  • Cheaper input tokens

DeepSeek V3.2 Speciale wins:

  • Cheaper output tokens
  • Higher intelligence benchmark
  • Better at coding
  • Better at math
Price Advantage
DeepSeek V3.1 Terminus
Benchmark Advantage
DeepSeek V3.2 Speciale
Context Window
DeepSeek V3.2 Speciale
Speed
N/A

Pricing Comparison

Price Comparison

MetricDeepSeek V3.1 TerminusDeepSeek V3.2 SpecialeWinner
Input (per 1M tokens)$0.21$0.27 DeepSeek V3.1 Terminus
Output (per 1M tokens)$0.79$0.41 DeepSeek V3.2 Speciale
Cache Read (per 1M)$130000.00$135000.00 DeepSeek V3.1 Terminus
Using a 3:1 input/output ratio, DeepSeek V3.2 Speciale is 14% cheaper overall.

DeepSeek V3.1 Terminus Providers

DeepInfra $0.21 (Cheapest)
Chutes $0.23
SiliconFlow $0.27
Novita $0.27
AtlasCloud $0.30

DeepSeek V3.2 Speciale Providers

Chutes $0.27 (Cheapest)
Parasail $0.40
AtlasCloud $0.40

Benchmark Comparison

6
Benchmarks Compared
0
DeepSeek V3.1 Terminus Wins
6
DeepSeek V3.2 Speciale Wins

Benchmark Scores

BenchmarkDeepSeek V3.1 TerminusDeepSeek V3.2 SpecialeWinner
Intelligence Index
Overall intelligence score
28.434.1
Coding Index
Code generation & understanding
31.937.9
Math Index
Mathematical reasoning
53.796.7
MMLU Pro
Academic knowledge
83.686.3
GPQA
Graduate-level science
75.187.1
LiveCodeBench
Competitive programming
52.989.6
DeepSeek V3.2 Speciale significantly outperforms in coding benchmarks.

Cost vs Quality

X-axis:
Y-axis:
Loading chart...
DeepSeek V3.1 Terminus
Other models

Context & Performance

Context Window

DeepSeek V3.1 Terminus
163,840
tokens
DeepSeek V3.2 Speciale
163,840
tokens
Max output: 65,536 tokens

Speed Performance

MetricDeepSeek V3.1 TerminusDeepSeek V3.2 SpecialeWinner
Tokens/second0.0 tok/s0.0 tok/s
Time to First Token0.00s0.00s

Capabilities

Feature Comparison

FeatureDeepSeek V3.1 TerminusDeepSeek V3.2 Speciale
Vision (Image Input)
Tool/Function Calls
Reasoning Mode
Audio Input
Audio Output
PDF Input
Prompt Caching
Web Search

License & Release

PropertyDeepSeek V3.1 TerminusDeepSeek V3.2 Speciale
LicenseOpen SourceOpen Source
AuthorDeepseekDeepseek
ReleasedSep 2025Dec 2025

DeepSeek V3.1 Terminus Modalities

Input
text
Output
text

DeepSeek V3.2 Speciale Modalities

Input
text
Output
text

Related Comparisons

Compare DeepSeek V3.1 Terminus with:

Compare DeepSeek V3.2 Speciale with:

Frequently Asked Questions

DeepSeek V3.1 Terminus has cheaper input pricing at $0.21/M tokens. DeepSeek V3.2 Speciale has cheaper output pricing at $0.41/M tokens.
DeepSeek V3.2 Speciale scores higher on coding benchmarks with a score of 37.9, compared to DeepSeek V3.1 Terminus's score of 31.9.
DeepSeek V3.1 Terminus has a 163,840 token context window, while DeepSeek V3.2 Speciale has a 163,840 token context window.
DeepSeek V3.1 Terminus does not support vision. DeepSeek V3.2 Speciale does not support vision.