Llama 3.1 Nemotron Ultra 253B V1 (Reasoning)

A 253B parameter derivative of Meta Llama 3.1 405B Instruct, developed by NVIDIA using Neural Architecture Search (NAS) and vertical compression. It has undergone multi - stage post - training (SFT for math, code, reasoning, chat, and tool invocation; RL with GRPO) to enhance reasoning and instruction - following capabilities. Optimized for the precision/efficiency trade - off on NVIDIA GPUs. Supports a 128k context.

Intelligence(Medium)

Speed(Slow)

Input Supported Modalities

Is Reasoning Model

128,000

Context Window

131,072

Maximum Output Tokens

2023-12-01

Knowledge Cutoff

Pricing

- /M tokens

Input

- /M tokens

Output

￥6.48 /M tokens

Blended Price

Quick Simple Comparison

Llama 3.3 Nemotron Super 49B v1 (Reasoning)

Llama 3.3 Nemotron Super 49B v1

Llama 3.1 Nemotron Ultra 253B v1 (Reasoning)

Basic Parameters

GPT-4.1 Technical Parameters

Parameter Count

253,000.0M

Context Length

128.00k tokens

Training Data Cutoff

2023-12-01

Open Source Category

Open Weights (Permissive License)

Multimodal Support

Text Only

Throughput

Release Date

2025-04-07

Response Speed

41.972,015 tokens/s

Benchmark Scores

Below is the performance of claude-monet in various standard benchmark tests. These tests evaluate the model's capabilities in different tasks and domains.

Intelligence Index

Large Language Model Intelligence Level

6082

Large Language Model Intelligence Level

Coding Index

Indicator of AI model performance on coding tasks

4942

Indicator of AI model performance on coding tasks

Math Index

Capability indicator in solving mathematical problems, mathematical reasoning, or performing math-related tasks

MMLU Pro

Massive Multitask Multimodal Understanding - Testing understanding of text, images, audio, and video

82.5

Massive Multitask Multimodal Understanding - Testing understanding of text, images, audio, and video

GPQA

Graduate Physics Questions Assessment - Testing advanced physics knowledge with diamond science-level questions

72.8

Graduate Physics Questions Assessment - Testing advanced physics knowledge with diamond science-level questions

HLE

The model's comprehensive average score on the Hugging Face Open LLM Leaderboard

8.1

The model's comprehensive average score on the Hugging Face Open LLM Leaderboard

LiveCodeBench

Specific evaluation focused on assessing large language models' ability in real-world code writing and solving programming competition problems

64.1

Specific evaluation focused on assessing large language models' ability in real-world code writing and solving programming competition problems

SciCode

The model's capability in code generation for scientific computing or specific scientific domains

34.7

The model's capability in code generation for scientific computing or specific scientific domains

HumanEval

Score achieved by the AI model on the specific HumanEval benchmark test set

Math 500 Score

Score on the first 500 larger, more well-known mathematical benchmark tests

95.2

Score on the first 500 larger, more well-known mathematical benchmark tests

AIME Score

An indicator measuring an AI model's ability to solve high-difficulty mathematical competition problems (specifically AIME level)

74.7

An indicator measuring an AI model's ability to solve high-difficulty mathematical competition problems (specifically AIME level)

Featured Recommended AI Models

Gemini 2.0 Flash Lite (Preview)

Output tokens/million

Output tokens/million

33k

Context Length

Qwen2.5 Coder Instruct 32B

Output tokens/million

Output tokens/million

Output tokens/million

Output tokens/million

128k

Context Length

Gemini 2.0 Pro Experimental (Feb '25)

google

Input tokens/million

Output tokens/million

Context Length

Llama 3.2 Instruct 11B (Vision)