LIVE INTELLIGENCE · part of the Xentir AI Index

DeepSeek-V3.2 (Thinking; Novita) vs GPT-5.4 nano

DeepSeek-V3.2 (Thinking; Novita) OPEN WEIGHTS · DeepSeek, released 2025-12-01
GPT-5.4 nano · OpenAI, released 2026-03-17

“Epoch AI, 'AI Benchmarking Hub'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks' [online resource].” · epoch.ai/benchmarks · Licensed CC BY 4.0 · data last verified 2026-07-26 · Methodology · Data Sources

Head-to-head

Measured on 3 of 9 tracked benchmarks for both models — GPT-5.4 nano leads on 3.

Benchmark	DeepSeek-V3.2 (Thinking; Novita)	GPT-5.4 nano	Leads
Epoch Capabilities Index (ECI)	146.6	146.6	GPT-5.4 nano
SWE-bench Verified	Not measured	Not measured	Not measured on both
GPQA Diamond	Not measured	0.785 (high)	Not measured on both
MATH Level 5	Not measured	Not measured	Not measured on both
OTIS Mock AIME 2024-2025	Not measured	0.878 (high)	Not measured on both
FrontierMath	0.221	0.259 (high)	GPT-5.4 nano
FrontierMath Tier 4	0.021	0.062 (high)	GPT-5.4 nano
SimpleQA Verified	Not measured	0.120 (high)	Not measured on both
Chess Puzzles	Not measured	0.300 (high)	Not measured on both

Benchmark results measure specific tasks and should not be treated as a universal measure of model quality. A model with no row here for a benchmark has not been measured on it — that is left blank, never estimated.

Full profiles

DeepSeek-V3.2 (Thinking; Novita) → · GPT-5.4 nano → · the full AI Model Rankings table