Nemotron-Cascade-14B-Thinking

Home/Models/nvidia/Nemotron-Cascade-14B-Thinking

nvidia/Nemotron-Cascade-14B-Thinking

kind: original
quant: bf16 (16-bit)
family: nemotron
modality: text-generation
root model: nvidia/Nemotron-Cascade-14B-Thinking
provider: nvidia
license: other
language: en
context: 90K
library: transformers
vRAM (GB): —
size on disk: —
downloads: 1,343
likes: 81
created: —
last modified: 2026-01-01T01:24:38+00:00 (2026)
huggingface: https://huggingface.co/nvidia/Nemotron-Cascade-14B-Thinking
tags: transformerssafetensorsqwen3text-generationnvidianemotron-cascadereasoninggeneral-purposeSFTRLpytorchconversationalenarxiv:2512.13607arxiv:2309.00071license:othertext-generation-inferenceendpoints_compatibleregion:us

benchmark	category	score	context	conf	evidence
Chatbot Arena	chat	91.7	ArenaHard	medium	`\| ArenaHard \| 91.7 \| 95.1 \| 95.7 \| 89.5 \|`
IFEval	instruction-following	85.4	IFEval (Strict Prompt)	medium	`\| IFEval (Strict Prompt) \| 85.4 \| 84.1 \| 89.8 \| 81.9 \|`
MMLU	knowledge	84.9	MMLU	medium	`\| MMLU \| 84.9 \| 89.9 \| - \| 85.1 \|`
AIME 2024	math	79.3	AIME 2024	medium	`\| AIME 2024 \| 79.3 \| 91.4 \| 82.3 \| 89.7 \|`
MMLU-Pro	knowledge	77.6	MMLU Pro	medium	`\| MMLU Pro \| 77.6 \| 85.0 \| 81.9 \| 77.0 \|`
AIME 2025	math	70.4	AIME 2025	medium	`\| AIME 2025 \| 70.4 \| 87.5 \| 72.0 \| 83.3 \|`
BFCL (function calling)	agentic	70.4	BFCL V3	medium	`\| BFCL V3 \| 70.4 \| 67.9 \| 68.6 \| 67.5 \|`
GPQA Diamond	reasoning	64.0	GPQA-Diamond	medium	`\| GPQA-Diamond \| 64.0 \| 81.0 \| 82.8 \| 69.6 \|`
IFBench	instruction-following	33.7	IFBench	medium	`\| IFBench \| 33.7 \| 38.0 \| 36.1 \| 41.7 \|`

none scraped.