Price Per TokenPrice Per Token
Deepseek
Deepseek
vs
Mistral AI
Mistral AI

DeepSeek V3.2 Speciale vs Devstral 2 2512

A detailed comparison of pricing, benchmarks, and capabilities

108 out of our 483 tracked models have had a price change in March.

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

Key Takeaways

DeepSeek V3.2 Speciale wins:

  • Higher intelligence benchmark
  • Better at coding
  • Better at math
  • Has reasoning mode

Devstral 2 2512 wins:

  • Cheaper output tokens
  • Larger context window
  • Faster response time
Price Advantage
Devstral 2 2512
Benchmark Advantage
DeepSeek V3.2 Speciale
Context Window
Devstral 2 2512
Speed
Devstral 2 2512

Pricing Comparison

Price Comparison

MetricDeepSeek V3.2 SpecialeDevstral 2 2512Winner
Input (per 1M tokens)$0.40$0.40Tie
Output (per 1M tokens)$1.20$0.90 Devstral 2 2512
Cache Read (per 1M)$0.20$0.45 DeepSeek V3.2 Speciale
Using a 3:1 input/output ratio, Devstral 2 2512 is 13% cheaper overall.

DeepSeek V3.2 Speciale Providers

No provider data available

Devstral 2 2512 Providers

No provider data available

Benchmark Comparison

6
Benchmarks Compared
6
DeepSeek V3.2 Speciale Wins
0
Devstral 2 2512 Wins

Benchmark Scores

BenchmarkDeepSeek V3.2 SpecialeDevstral 2 2512Winner
Intelligence Index
Overall intelligence score
29.422.0
Coding Index
Code generation & understanding
37.923.7
Math Index
Mathematical reasoning
96.736.7
MMLU Pro
Academic knowledge
86.376.2
GPQA
Graduate-level science
87.159.4
LiveCodeBench
Competitive programming
89.644.8
DeepSeek V3.2 Speciale significantly outperforms in coding benchmarks.

Cost vs Quality

X-axis:
Y-axis:
Loading chart...
DeepSeek V3.2 Speciale
Other models

Context & Performance

Context Window

DeepSeek V3.2 Speciale
163,840
tokens
Devstral 2 2512
262,144
tokens
Devstral 2 2512 has a 38% larger context window.

Speed Performance

MetricDeepSeek V3.2 SpecialeDevstral 2 2512Winner
Tokens/second0.0 tok/s81.2 tok/s
Time to First Token0.00s0.39s
Devstral 2 2512 responds Infinity% faster on average.

Capabilities

Feature Comparison

FeatureDeepSeek V3.2 SpecialeDevstral 2 2512
Vision (Image Input)
Tool/Function Calls
Reasoning Mode
Audio Input
Audio Output
PDF Input
Prompt Caching
Web Search

License & Release

PropertyDeepSeek V3.2 SpecialeDevstral 2 2512
LicenseOpen SourceOpen Source
AuthorDeepseekMistral AI
ReleasedDec 2025Dec 2025

DeepSeek V3.2 Speciale Modalities

Input
text
Output
text

Devstral 2 2512 Modalities

Input
text
Output
text

Related Comparisons

Compare DeepSeek V3.2 Speciale with:

Compare Devstral 2 2512 with:

Frequently Asked Questions

DeepSeek V3.2 Speciale has cheaper input pricing at $0.40/M tokens. Devstral 2 2512 has cheaper output pricing at $0.90/M tokens.
DeepSeek V3.2 Speciale scores higher on coding benchmarks with a score of 37.9, compared to Devstral 2 2512's score of 23.7.
DeepSeek V3.2 Speciale has a 163,840 token context window, while Devstral 2 2512 has a 262,144 token context window.
DeepSeek V3.2 Speciale does not support vision. Devstral 2 2512 does not support vision.