PipelineScore
← Back to leaderboard
meta

Llama 3.3 70B Instruct

Released 2024-12-06Context 128Kllama-3-3-70b-instruct
PipelineScore
78.7MAINLINE
Ranked #12 of 52 models · 77th percentileCode is the headline (80.7); throughput is the soft spot (71.8). Best-fit profile: Coding.

Category breakdown

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

Code
80.7
Reason
79.7
Tool Use
79.3
RAG
77.9
Speed
71.8

Strengths

Code80.7
Reason79.7
Tool Use79.3

Same model, different rigs

Every submission of Llama 3.3 70B Instruct on the 0–100 scale. The spread is the point: where it runs changes what you get.

0255075100

Best 83.3 on h100-80gb · lowest 76.9 on m3-ultra-256gb · spread 6.4 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

Llama 3.3 70B Instruct vs openai/gpt-oss-20bLlama 3.3 70B Instruct vs DeepSeek R1 671B-A37BLlama 3.3 70B Instruct vs Qwen 3 235B-A22B MoELlama 3.3 70B Instruct vs DeepSeek V3 671B-A37BLlama 3.3 70B Instruct vs DeepSeek Coder V2 236B