Model
Model explorer

Llama-3.1-Nemotron-70B Instruct

OPEN
NVIDIA · Llama Nemotron family · released Oct 15, 2024

RLHF (REINFORCE)-tuned Llama-3.1-70B that led Arena Hard/AlpacaEval/MT-Bench in Oct 2024, edging GPT-4o and Claude 3.5 Sonnet.

ReasoningCodingVisionFunction callingTool useAgentic
661.5
Elo · rank #272
Parameters
70.6B
Active params
70.6B (dense)
Context
128K tokens
Architecture
Dense Transformer (Llama-3.1-70B base, REINFORCE/RLHF-tuned)
License
NVIDIA Open Model License + Llama 3.1 Community License Agreement
Languages
—
API price (in/out)
$1.2 / $1.2
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
AlpacaEvalHuman preference57.6%#14
best: Tülu 2+DPO 70B · 95.1%
Arena-HardHuman preference85.0%#22
best: Qwen3-235B-A22B (Non-Thinking) · 96.1%
Arena EloHuman preference1267#36
best: Claude Fable 5 · 1505
BIG-Bench HardReasoning63.0%#75
best: ERNIE 4.5 300B-A47B · 94.3%
GPQA DiamondReasoning25.8%#276
best: GPT-5.6 · 94.6%
IFEvalReasoning78.3%#95
best: Gemma 4 26B A4B · 98.5%
MATH-500Math27.0%#169
best: GPT-5 · 99.4%
MMLU-ProKnowledge62.8%#112
best: Claude Fable 5 · 91.5%
MT-BenchHuman preference8.98#11
best: Hunyuan-Large (A52B) · 9.4
Run it locally
VRAM @ Q4
42 GB
VRAM @ FP16
145 GB
Fits on (Q4)
M3 Max 128GBM3 Ultra 512GBA100 80GBH100 80GBH200 141GBB200 192GB
Needs ~42GB at Q4 — doesn't fit a single 24GB RTX 4090; no verified 4090 benchmark found.
Quantizations
GGUF · AWQ
Llama Nemotron family
Elo progression across releases
Llama-3.1-Nemotron-70B InstructOct 2024661.5Llama-3.3-Nemotron-Super-49B v1 (Reasoning Off)Mar 20251116.2Llama-3.3-Nemotron-Super-49B v1 (Reasoning On)Mar 20251378.0Llama-3.1-Nemotron-Ultra-253B v1 (Reasoning Off)Apr 20251120.4Llama-3.1-Nemotron-Ultra-253B v1 (Reasoning On)Apr 20251544.6
Resources
HFnvidia/Llama-3.1-Nemotron-70B-Instruct-HFDOCSdocs.api.nvidia.com/nim/reference/nvidia-llama-3_1-nemotron-70b-instructAPIAvailable via NVIDIA API
API price $1.2/$1.2 · each benchmark row carries its own source badge (see methodology)