Inference
EN
Meta

Llama 3.1 70B

Llama 3.1 70B is a standard model from Meta. 14 benchmarks count toward its index, in 7 categories.

  • Open weights
  • Standard
  • Input: text
  • Released 2024-07
RankedEnough results for an index.
IndexAll categories, out of 100.24.8±5.3Median 51.0
CoverageIndex weight that has results.90%
SpeedOutput tokens per second.—Median 67/s
First tokenSeconds to the first output token.—Median 1.3 s
Input / 1MUS dollars per 1M input tokens.—Median $0.60
Output / 1MUS dollars per 1M output tokens.—Median $2.55
ContextMaximum tokens in one request.128KMedian 262K
EloLMArena rating and rank.—Median 1422

Run locally

70.55B parameters
16-bit
149.2 GB
8-bit
79.7 GB
4-bit
45.9 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy fits in 48 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.1 counted17.1—
CodingCode writing and repair.1 counted21.4—
ReasoningLogic problems and puzzles.2 counted23.7220 / 227
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.4 counted27.6247 / 268
MultilingualTasks in many languages.2 counted41.07 / 9
InstructionTasks with strict prompt rules.2 counted27.857 / 63
MathMath problems.2 counted13.9166 / 168

Results

14 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
EuroEval PolishMultilingualAbove fitFar above this model's index.56.5% ±0.532 / 209
EuroEval GermanMultilingualAbove fitFar above this model's index.52.1% ±0.630 / 191
EuroEval PortugueseMultilingualAbove fitFar above this model's index.59.3% ±0.432 / 182
EuroEval FrenchMultilingualAbove fitFar above this model's index.58.4% ±0.545 / 178
Multilingual Grade School MathMultilingual86.9%4 / 13
Massive Multitask Language UnderstandingKnowledge86.0%16 / 34
Instruction-Following EvalInstruction87.5%33 / 70
Evaluating Large Language Models Trained on CodeCoding80.5%15 / 22
Critical Physics TasksReasoning0.0%150 / 214
MATH level 5Math36.7% ±1.062 / 88
GPQA diamond · Epoch AIKnowledge44.2% ±2.5166 / 217
Massive Multitask Language Understanding ProfessionalKnowledge66.4%62 / 75
τ²-Bench Tool-Agent-User EvaluationAgentic21.6%127 / 149
OTIS Mock AIME 2024-2025Math3.6% ±1.7176 / 200
Artificial Analysis GPQA DiamondKnowledge49.1%176 / 199
Graduate-Level Google-Proof Q&AKnowledge46.7%126 / 141
Artificial Analysis IFBenchInstruction29.0%136 / 146
Artificial Analysis Long Context ReasoningReasoning4.7%203 / 217
Artificial Analysis Humanity's Last ExamKnowledge3.6%210 / 222
Grade School Math 8KMath95.1%—

Sources

BenchLMCC BY-NC 4.0EuroEvalMITEpoch AICC BY

History

5 releases
Index27.2 → 24.8
23 Oct 202612 Nov 2026

Questions

How good is Llama 3.1 70B?
Its index is 24.8 out of 100. That is place 269 of 283 ranked models.
What is Llama 3.1 70B best at?
Multilingual: 41.0 out of 100, place 7 of 9.
What is the context window of Llama 3.1 70B?
128K tokens.
Is Llama 3.1 70B open source?
Yes. Its weights are published.
When was Llama 3.1 70B released?
July 2024.

Badge

Llama 3.1 70B on Inference360: place 269 of 283, index 24.8
HTML
<a href="https://inference360.ai/models/llama-3-1-70b"><img src="https://inference360.ai/badges/llama-3-1-70b.svg" alt="Llama 3.1 70B on Inference360: place 269 of 283, index 24.8" height="20"></a>
Markdown
[![Llama 3.1 70B on Inference360: place 269 of 283, index 24.8](https://inference360.ai/badges/llama-3-1-70b.svg)](https://inference360.ai/models/llama-3-1-70b)

Compare with

Gemma 3 27B25.3Mistral Small 3.124.2Nova Pro25.6Llama 3.1 405B27.0Llama 3.1 8B17.9

More from Meta

All models
Llama 3.1 405B27.0Llama 3.1 8B17.9Llama 3.1 70B InstructUnrankedLlama 3.1 8B InstructUnrankedLlama 3.2 1B InstructUnranked