Qwen

Qwen 2.5 72B — Benchmark Scores, Pricing & Performance Analysis

OPEN SOURCEQwen
Chatbot Arena ELO
1230
Output Speed
80 tok/s
Input Cost
$0.36/1M
Output Cost
$0.40/1M
Context Window
33K
Max Output
16K
Knowledge Cutoff
Jun 2024
Accepts
text

Qwen 2.5 72B by Qwen demonstrates competitive pricing. View detailed benchmark data including scores across coding, math, reasoning, speed, and cost metrics.

Specifications last checked against the provider on . Benchmark scores carry their own source, linked beside each number.

General Benchmarks

MMLU-Pro
72.0%42nd of 66

Massive Multitask Language Understanding professional benchmark

HuggingFace Leaderboard
Chatbot Arena ELO
123062nd of 74

Human preference ELO from blind head-to-head votes

LMSYS / HuggingFace
TruthfulQA
74.0%12th of 13

Resistance to generating false but plausible answers

Papers
AlpacaEval 2.0
49.1%8th of 10

Instruction following quality scored by GPT-4 as judge

tatsu-lab

Coding Benchmarks

LiveCodeBench
55.5%22nd of 66

Live competitive programming benchmark

livecodebench.github.io
SWE-bench Verified
28.0%51st of 67

Real-world software engineering task resolution

swebench.com
HumanEval+
80.0%43rd of 66

Code generation correctness with extended tests

Papers

Math Benchmarks

GSM8K
91.0%38th of 66

Grade school math word problems

Papers
MATH
83.1%30th of 66

Competition mathematics problem solving

Papers

Reasoning Benchmarks

ARC-AGI
12.0%55th of 66

Abstraction and Reasoning Corpus for general intelligence

arcprize.org
GPQA Diamond
49.1%116th of 140

Graduate-level science Q&A by domain experts

Papers
BBH
84.0%11th of 13

Big-Bench Hard — 23 challenging multi-step reasoning tasks

Google Research
Winogrande
86.0%13th of 14

Commonsense reasoning via pronoun resolution

Papers

Speed Benchmarks

Time to First Token
350 ms31st of 92

Latency before first token arrives

Aggregated
Output Speed
80 tok/s45th of 94

Tokens generated per second

Aggregated

Cost Benchmarks

Input Cost
$0.36225th of 413

Cost per 1M input tokens

OpenRouter API
Output Cost
$0.40143rd of 413

Cost per 1M output tokens

OpenRouter API

Context Benchmarks

Context Length
33K369th of 451

Maximum context window size

OpenRouter API
RULER
85.9%7th of 13

Long-context understanding and retrieval accuracy at depth

Papers

Multimodal Benchmarks

MathVista
67.5%10th of 14

Visual mathematical reasoning across diagrams and charts

Papers

Qwen 2.5 72B — Benchmark Scores Overview

Scores normalized to percentage scale for visual comparison. ELO scores mapped to 0-100 range (1100-1500).

Qwen 2.5 72B — Frequently Asked Questions

How intelligent is Qwen 2.5 72B?

Qwen 2.5 72B scores 1230 on the Chatbot Arena ELO rating, making it an entry-level AI model. This score is based on blind head-to-head human preference voting.

How much does Qwen 2.5 72B cost?

Qwen 2.5 72B costs $0.36 per 1M input tokens and $0.40 per 1M output tokens. This makes it one of the more affordable models.

How fast is Qwen 2.5 72B?

Qwen 2.5 72B generates output at 80 tokens per second, which is moderate compared to other models. The time to first token is 350 ms.

How good is Qwen 2.5 72B at coding?

Qwen 2.5 72B achieves 28.0% on SWE-bench Verified, demonstrating basic real-world software engineering capability. This benchmark tests the model's ability to resolve actual GitHub issues.

How good is Qwen 2.5 72B at math and reasoning?

Qwen 2.5 72B scores 83.1% on the MATH benchmark (competition-level mathematics). It also achieves 49.1% on GPQA Diamond, a graduate-level science reasoning benchmark.

What is the context window of Qwen 2.5 72B?

Qwen 2.5 72B has a context window of 33K tokens. This determines how much text, conversation history, and code the model can process in a single request.

Who created Qwen 2.5 72B?

Qwen 2.5 72B was created by Qwen. It is classified as a open source model in our catalogue.

Is Qwen 2.5 72B open source?

Yes, Qwen 2.5 72B is an open-source model. The model weights are publicly available for download and self-hosting.