DeepSeek-V3.2
DeepSeek
Parameters
—
Architecture
—
Released
—
License
—
About
No detailed description available for this model.
Benchmark Scores
| Benchmark | Score | Date |
|---|---|---|
|
SWE-bench Verified
coding_agent
|
80.05%
|
03.02.2026 |
|
SWE-bench Multilingual
coding_agent
|
67.81%
|
03.02.2026 |
|
SWE-bench Pro
coding_agent
|
51.12%
|
03.02.2026 |
|
Terminal-Bench 2.0
coding_agent
|
20.22%
|
03.02.2026 |
|
Aider
coding_agent
|
100.00%
|
03.02.2026 |
|
MMLU
knowledge
|
92.46%
|
26.06.2026 |
|
MMLU-Redux
knowledge
|
65.97%
|
26.06.2026 |
|
MMLU-Pro
knowledge
|
17.74%
|
26.06.2026 |
|
MMMLU
multilingual
|
89.52%
|
26.06.2026 |
|
C-Eval
knowledge
|
75.23%
|
26.06.2026 |
|
SuperGPQA
knowledge
|
42.34%
|
26.06.2026 |
|
LongBench v2
long_context
|
34.62%
|
26.06.2026 |