DeepSeek-V3.2-Exp

Home/Models/deepseek-ai/DeepSeek-V3.2-Exp

deepseek-ai/DeepSeek-V3.2-Exp

kind: fp8
quant: fp8 (8-bit)
family: deepseek
modality: text-generation
root model: deepseek-ai/DeepSeek-V3.2-Exp-Base
provider: deepseek-ai
license: mit
language: —
context: —
library: transformers
vRAM (GB): —
size on disk: —
downloads: 291,271
likes: 992
created: —
last modified: 2025-11-18T02:39:54+00:00 (2025)
huggingface: https://huggingface.co/deepseek-ai/DeepSeek-V3.2-Exp
tags: transformerssafetensorsdeepseek_v32text-generationconversationalbase_model:deepseek-ai/DeepSeek-V3.2-Exp-Basebase_model:finetune:deepseek-ai/DeepSeek-V3.2-Exp-Baselicense:miteval-resultsendpoints_compatiblefp8region:us

benchmark	category	score	context	conf	evidence
SimpleQA	knowledge	96.8	SimpleQA	medium	`\| SimpleQA \| 96.8 \| 97.1 \|`
AIME 2025	math	88.4	AIME 2025	medium	`\| AIME 2025 \| 88.4 \| 89.3 \|`
MMLU-Pro	knowledge	85.0	MMLU-Pro	medium	`\| MMLU-Pro \| 85.0 \| 85.0 \|`
GPQA Diamond	reasoning	80.7	GPQA-Diamond	medium	`\| GPQA-Diamond \| 80.7 \| 79.9 \|`
LiveCodeBench	coding	74.9	LiveCodeBench	medium	`\| LiveCodeBench \| 74.9 \| 74.1 \|`
BrowseComp	agentic	45.0	BrowseComp-zh	medium	`\| BrowseComp-zh \| 45.0 \| 47.9 \|`
BrowseComp	agentic	38.5	BrowseComp	medium	`\| BrowseComp \| 38.5 \| 40.1 \|`
Terminal-Bench	agentic-coding	36.7	Terminal-bench	medium	`\| Terminal-bench \| 36.7 \| 37.7 \|`
Humanity's Last Exam	knowledge	21.7	Humanity's Last Exam	medium	`\| Humanity's Last Exam \| 21.7 \| 19.8 \|`

none scraped.