Meta Llama 3 8B Instruct — Benchmarks

Benchmark scores for Meta Llama 3 8B Instruct aggregated from public leaderboards (12 boards), with how it ranks among open models and how it compares with its nearest neighbours. See hardware requirements for what you need to run it.

llmrun Composite Score →

61.7

90% range 54.0–69.4 · from 5 boards

CodingNot enough data yet
AgentsNot enough data yet
MathNot enough data yet
ScienceNot enough data yet
Reasoning63.990% range 49–79

Meta Llama 3 8B Instruct reasoning benchmarks

Multi-step logic and problem-solving — can the model think through a hard problem rather than recall an answer.

GPQA Diamond
GPQA Diamond is a set of extremely hard, graduate-level science questions (physics, chemistry, biology) written by domain experts and filtered so that skilled non-experts with web access still fail. It measures genuine reasoning rather than memorization.
26.1%
26.1%
median 54.0%open leader 93.1%
Overall best: Gemini 4 Argon (high) 96.4%
9th percentile · #89 of 98 open · #289 of 300 all
#89 of 98
#289 of 300
Independent
Epoch AI · Apr 2024
DTBench
DTBench is a set of handcrafted, expert-validated multiple-choice questions on decision theory — Newcomb-like problems and related puzzles — built by the Conceptual Reasoning Index project with Anthropic. It tests whether a model can follow an abstract argument rather than recall a fact.
43.9%
43.9%
median 67.7%open leader 93.9%
Overall best: Claude Opus 5.5 Max 98.9%
3rd percentile · #66 of 68 open · #211 of 214 all
#66 of 68
#211 of 214
Lab-reported
Epoch AI · Apr 2024
ForecastBench
ForecastBench, from the Forecasting Research Institute, asks models to predict real-world future events, with questions that only resolve after the model's training cutoff so the answers can't have been memorised. It is scored as a difficulty-adjusted Brier index, where higher is better.
58.6
58.6
median 58.1open leader 61.9
Overall best: O3 (Apr 16, 2025, unspecified) 62.5
58th percentile · #12 of 26 open · #53 of 82 all
#12 of 26
#53 of 82
Lab-reported
Epoch AI · Apr 2024

Other results (1)· leaderboard-only, not part of the llmrun Score

WinoGrande
WinoGrande is a large-scale commonsense-reasoning test built around pronoun resolution: the model must pick which of two nouns a pronoun refers to in a sentence, using real-world knowledge to disambiguate. It was designed by AI2 to resist the shortcuts that let earlier models game its predecessor, Winograd.
65.0%
65.0%
median 70.1%open leader 89.2%
34th percentile · #34 of 51 open · #61 of 79 all
#34 of 51
#61 of 79
Lab-reported
Epoch AI · Apr 2024

Meta Llama 3 8B Instruct math benchmarks

Symbolic and quantitative problem-solving, from competition math to multi-step calculation.

AIME 2024/2025
AIME (American Invitational Mathematics Examination) is a prestigious high-school competition of hard, integer-answer problems. It's a widely-cited yardstick for multi-step mathematical reasoning, here on the 2024–2025 papers.
1.9%
1.9%
median 52.8%open leader 98.6%
Overall best: Claude Fable 5 (high) 100.0%
12th percentile · #76 of 86 open · #268 of 283 all
#76 of 86
#268 of 283
Independent
Epoch AI · Apr 2024
MATH Level 5
MATH Level 5 covers the hardest tier of competition-mathematics problems, testing multi-step symbolic and quantitative reasoning.
6.1%
6.1%
median 41.6%open leader 96.6%
Overall best: GPT 5 (Aug 07, 2025, high) 98.1%
8th percentile · #36 of 39 open · #103 of 107 all
#36 of 39
#103 of 107
Independent
Epoch AI · Apr 2024

Meta Llama 3 8B Instruct knowledge benchmarks

Breadth and depth of factual knowledge across many subjects, paired with the reasoning to apply it.

MMLU-Pro
MMLU-Pro is a harder, cleaned-up successor to MMLU with ten answer choices and more reasoning-heavy questions across 14 subjects, measuring broad knowledge and reasoning together.
41.0%
41.0%
median 52.1%open leader 88.0%
Overall best: Gemini-3.1-Pro 91.2%
31st percentile · #100 of 145 open · #195 of 257 all
#100 of 145
#195 of 257
Independent
TIGER-Lab

Other results (4)· leaderboard-only, not part of the llmrun Score

MMLU
MMLU (Massive Multitask Language Understanding) spans 57 subjects from history to law to medicine as multiple-choice questions. It is the long-standing default for broad knowledge and remains the most widely-reported general benchmark.
68.8%
68.8%
median 62.5%open leader 87.2%
Overall best: GPT 4o (Nov 20, 2024) 88.1%
65th percentile · #30 of 85 open · #66 of 136 all
#30 of 85
#66 of 136
Lab-reported
Epoch AI · Apr 2024
ARC Challenge
ARC Challenge is a set of grade-school science questions, curated by the Allen Institute for AI (AI2) to exclude ones that simple word-matching or retrieval can answer. It measures science knowledge paired with the reasoning needed to apply it.
82.8%
82.8%
median 53.9%open leader 95.3%
82nd percentile · #11 of 56 open · #16 of 76 all
#11 of 56
#16 of 76
Lab-reported
Epoch AI · Apr 2024
OpenBookQA
OpenBookQA is a science-question benchmark modeled on open-book exams: the model is given a small set of core science facts and must combine them with broader common knowledge to answer multi-step questions. It targets reasoning over recall.
82.6%
82.6%
median 56.8%open leader 88.0%
84th percentile · #5 of 26 open · #6 of 41 all
#5 of 26
#6 of 41
Lab-reported
Epoch AI · Apr 2024
TriviaQA
TriviaQA is a large set of trivia questions paired with evidence documents, testing a model's factual recall and reading comprehension across a wide range of everyday topics.
67.7%
67.7%
median 73.7%open leader 87.6%
28th percentile · #14 of 19 open · #31 of 39 all
#14 of 19
#31 of 39
Lab-reported
Epoch AI · Apr 2024

Meta Llama 3 8B Instruct instruction following benchmarks

How precisely a model obeys explicit, verifiable instructions — format, length, and content constraints — and, more broadly, how much people prefer its responses head-to-head in blind human voting.

LMArena Text
Arena (arena.ai, formerly LMArena / Chatbot Arena) ranks models by millions of blind, head-to-head human votes on real prompts. The Arena Score shown here is a Bradley-Terry rating computed with Arena's default style control, so higher means people preferred that model's answers more often — it's a relative rating, not a percentage, so compare it only to other models on this board.
1223.7
1223.7
median 1321.5open leader 1487.1
Overall best: Gemini 4 Argon (High) 1523.9
27th percentile · #137 of 187 open · #327 of 405 all
#137 of 187
#327 of 405
Arena
Arena · Oct 2026

Score vs VRAM in the 24 GB tier

Meta Llama 3 8B Instruct
Other models
Best Score for its VRAM
x: VRAM (GB) · y: llmrun Score

#35 of 35 that fit 24 GB

Each dot is a model at the best quant that fits a 24 GB GPU. The line joins the models with the best Score for their VRAM; a dot below it is beaten by a model that needs no more memory.

Full 24 GB ranking →

Meta Llama 3 8B Instruct vs its neighbours

Same boards for the nearest open models per llmrun Score and the best model that fits 24 GB. The highest score in each row is marked.

Board	Meta Llama 3 8B Instructthis model	Qwen3.8 27Bbest in classQ4_K_M · 18 GB	Qwen2.5 7B Instructnearest by ScoreQ4_K_M · 5.8 GB	Gemma 3 4B ITnearest by ScoreQ4_K_M · 5.0 GB	Llama 3.1 8B Instructnearest by ScoreQ4_K_M · 7.9 GB
llmrun Score	61.7	122.1	65.1	64.0	62.5
GPQA Diamond	26.1%	—	35.5%	23.2%	27.0%
DTBench	43.9%	88.0%	47.7%	50.9%	50.9%
AIME 2024/2025	1.9%	—	2.5%	7.5%	1.7%
MATH Level 5	6.1%	—	—	—	22.9%
MMLU-Pro	41.0%	—	—	43.6%	44.3%
LMArena Text	1223.7	1437.8	—	1303.3	1211.2
Compare head to head		Compare	Compare	Compare	Compare

Scores aggregated from public benchmark sources (Epoch AI: 10 boards, Arena: 1 board, TIGER-Lab: 1 board) — each linked from the benchmark pages. llmrun does not run these benchmarks. Percentiles and ranks are among open models in the llmrun catalogue; the median and leader marks refer to the same field.