LLM

DeepSeek V3.2 Exp (Reasoning)

Name: DeepSeek V3.2 Exp (Reasoning) Review & Benchmarks
Item: DeepSeek V3.2 Exp (Reasoning)
Rating: 16.6
Author: Sentient Weekly

Overall Score

16.6

Released Sep 2025

Benchmark Scores

Reasoning

—

Coding

33.3

Math

87.7

Creative writing

—

Instruction following

—

Multimodal

—

Standard Benchmarks

aa_mmlu_proaa_mmlu_pro

0.8

Measured 2026-09-26 · source

aa_gpqaaa_gpqa

0.8

Measured 2026-09-26 · source

aa_hleaa_hle

0.1

Measured 2026-09-26 · source

aa_scicodeaa_scicode

0.4

Measured 2026-09-04 · source

aa_livecodebenchaa_livecodebench

0.8

Measured 2026-09-26 · source

aa_aime_25aa_aime_25

0.9

Measured 2026-09-26 · source

aa_ifbenchaa_ifbench

0.5

Measured 2026-09-26 · source

aa_lcraa_lcr

0.7

Measured 2026-09-26 · source

aa_terminalbench_hardaa_terminalbench_hard

0.3

Measured 2026-09-26 · source

aa_tau2aa_tau2

0.3

Measured 2026-09-26 · source

Price

—

Context

—

Speed

0 t/s

0ms TTFT

Compare this model →

Discussion

0 comments

Be the first to comment.