PipelineScore
← Back to leaderboard
alibaba

Qwen 2.5 72B Instruct

Released 2024-09-19Context 131.072Kqwen-2-5-72b-instruct
PipelineScore
82.7MAINLINE
Ranked #6 of 52 models · 88th percentileAn even spread: no standout, no liability (81 to 86 across all five categories). Best-fit profile: Agentic.

Category breakdown

Score per category, normalized 0–100 against the v1 anchor.

Code
81.0
Reason
83.8
Tool Use
85.7
RAG
82.7
Speed
82.2

Strengths

Tool Use85.7
Reason83.8
RAG82.7

Same model, different rigs

Every submission of Qwen 2.5 72B Instruct on the 0–100 scale. The spread is the point: where it runs changes what you get.

0255075100

Best 85.1 on h200-141gb · lowest 79.9 on lab-h100-80gb · spread 5.2 pts across 4 runs. Hover a dot for its rig.

Sample tasks

A taste of what the test pack measures. Full prompts are private and rotated daily.

CodeDifficulty 1code-fib-1

Fibonacci function

Write a Python `fib(n)` returning the nth Fibonacci number, O(n).

ReasonDifficulty 1reason-math-1

Train meeting time

Two trains, opposite directions, given speeds and start times — when do they meet?

RAGDifficulty 2rag-extract-1

Extract metrics to JSON

From the context, extract net sales, operating margin, and free cash flow as a JSON object. Numbers only.

Tool UseDifficulty 2tool-schema-1

OpenAPI param selection

Given an OpenAPI schema with limit/offset/sort, fill JSON for 'next 50, recent first.'

RAGDifficulty 2rag-grounding-1

Refuses to fabricate

Context lacks the answer — does the model fabricate or correctly say it can't?

Compare with

Qwen 2.5 72B Instruct vs openai/gpt-oss-20bQwen 2.5 72B Instruct vs DeepSeek R1 671B-A37BQwen 2.5 72B Instruct vs Qwen 3 235B-A22B MoEQwen 2.5 72B Instruct vs DeepSeek V3 671B-A37BQwen 2.5 72B Instruct vs DeepSeek Coder V2 236B