DeepSeek-V3.2-Exp

Training compute

4.2×10²⁴ FLOP

Parameters

671B

Published

Sep 29, 2025

DeepSeek-V3.2-Exp is an AI model developed by DeepSeek (China), first published in September 2025. It works in the language domain, on tasks such as language modeling/generation, code generation, quantitative reasoning and 4 more.

Training it took an estimated 4.2×10²⁴ FLOP of compute (estimation method: operation counting). The model has 671,000,000,000 parameters. It was trained on roughly 943.7B datapoints.

Access: Open weights (unrestricted). Its weights are openly available. It is built on top of DeepSeek-V3.1-Terminus. Epoch AI rates the confidence of this record as confident.

Full record

Organization: DeepSeek
Country of organization: China
Domain: Language
Task: Language modeling/generation, Code generation, Quantitative reasoning, Question answering, Search, System control, Instruction interpretation
Training compute: 4.2×10²⁴ FLOP
Compute estimation method: Operation counting
Parameters: 671,000,000,000
Dataset size: 943.7B
Model accessibility: Open weights (unrestricted)
Open weights: Yes
Base model: DeepSeek-V3.1-Terminus
Epoch confidence: Confident

Benchmark results

01	WebDev Arena	Arena Score	1290
02	Epoch Capabilities Index	ECI Score	144.9
03	Aider Polyglot	Percent correct	70.2
04	TheAgentCompany	% Score	0.52