DeepSeek-V3.2 (Thinking) vs DeepSeek-V4-Pro-0813
DeepSeek-V4-Pro-0813 significantly outperforms across most benchmarks. DeepSeek-V3.2 (Thinking) is 1.7x cheaper per token.
DeepSeek · DeepSeek · Updated for 2026
Which is better?
DeepSeek-V3.2 (Thinking) outperforms in 0 benchmarks, while DeepSeek-V4-Pro-0813 is better at 2 benchmarks (Humanity's Last Exam, Toolathlon). DeepSeek-V4-Pro-0813 significantly outperforms across most benchmarks.
On price, DeepSeek-V3.2 (Thinking) is roughly 1.7x cheaper per token on a blended 3:1 input/output basis, which adds up quickly at production volume.
DeepSeek-V4-Pro-0813 also accepts a larger context window (1,048,576 input tokens), making it the stronger choice for long documents and large codebases.
Based on current benchmark, pricing, and model metadata for 2026.
Choose DeepSeek-V3.2 (Thinking)
- cost matters — it's about 1.7x cheaper per token
Choose DeepSeek-V4-Pro-0813
- you want the strongest raw capability — it leads on 2 of 2 shared benchmarks
- you process long inputs — it offers a 1,048,576 token context window
- you want the most recent training data — it shipped Aug 2026
At a glance
The differences that matter most.
Performance Benchmarks
Comparative analysis across standard metrics
DeepSeek-V3.2 (Thinking) outperforms in 0 benchmarks, while DeepSeek-V4-Pro-0813 is better at 2 benchmarks (Humanity's Last Exam, Toolathlon).
DeepSeek-V4-Pro-0813 significantly outperforms across most benchmarks.
Arena Performance
Playground indexes and blind preference scores
Pricing Analysis
Price comparison per million tokens
For input processing, DeepSeek-V3.2 (Thinking) ($0.28/1M tokens) is 1.6x cheaper than DeepSeek-V4-Pro-0813 ($0.43/1M tokens).
For output processing, DeepSeek-V3.2 (Thinking) ($0.42/1M tokens) is 2.1x cheaper than DeepSeek-V4-Pro-0813 ($0.87/1M tokens).
In conclusion, DeepSeek-V4-Pro-0813 is more expensive than DeepSeek-V3.2 (Thinking).*
* Using a 3:1 ratio of input to output tokens
Model Size
Parameter count comparison
DeepSeek-V4-Pro-0813 has 915.0B more parameters than DeepSeek-V3.2 (Thinking), making it 133.6% larger.
Context Window
Maximum input and output token capacity
DeepSeek-V4-Pro-0813 accepts 1,048,576 input tokens compared to DeepSeek-V3.2 (Thinking)'s 131,072 tokens. DeepSeek-V4-Pro-0813 can generate longer responses up to 393,216 tokens, while DeepSeek-V3.2 (Thinking) is limited to 65,536 tokens.
License
Usage and distribution terms
Both models are licensed under MIT.
Both models share the same licensing terms, providing consistent usage rights.
MIT
Open weights
MIT
Open weights
Release Timeline
When each model was launched
DeepSeek-V3.2 (Thinking) was released on 2025-12-01, while DeepSeek-V4-Pro-0813 was released on 2026-08-13.
DeepSeek-V4-Pro-0813 is 9 months newer than DeepSeek-V3.2 (Thinking).
Dec 1, 2025
8 months ago
Aug 13, 2026
1 weeks ago
8mo newerKnowledge Cutoff
When training data ends
Neither model specifies a knowledge cutoff date.
Unable to compare the recency of their training data.
Provider Availability
DeepSeek-V3.2 (Thinking) is available from DeepSeek. DeepSeek-V4-Pro-0813 is available from DeepSeek, DeepInfra, Novita, Together.
DeepSeek-V3.2 (Thinking)
DeepSeek-V4-Pro-0813
Outputs Comparison
Judge for yourself.
Run your own prompts against DeepSeek-V3.2 (Thinking) and DeepSeek-V4-Pro-0813 side-by-side, then vote on the output you prefer.
FAQ
Common questions about DeepSeek-V3.2 (Thinking) vs DeepSeek-V4-Pro-0813.