Meta AI

Llama 3.1 405B

Released Jul 23, 2024 · knowledge cutoff 2023-12

Status: Deprecated
Location: United States
Type: LLM
Modality: Text
Context window: 128K
Max output: 4K
Speed: —
Price ($/MTok in / out): $3.50 / $3.50
Cost per task: —
Open weights: Yes

Benchmarks4 of 10 reported

MMLU-Proretired

73.3%

GPQA Diamond

51.1%

SWE-bench Pro

11.18%

LMArena Elo

1269

SWE-bench Verified

—

Terminal-Bench 2.1

—

AIME (latest)retired

—

Humanity's Last Exam

—

GDPval-AA v2

—

ARC-AGI-2

—

Strengths & weaknesses

Strengths

Runs entirely on your own hardware with no vendor in the loop
Fine-tunes cleanly — the base most of the open ecosystem was built on
Solid across major European languages rather than English with translations bolted on

Weaknesses

Needs a multi-GPU node to serve; not a single-machine model
Text-only, so no screenshots, charts or documents
One-pass answers with no way to make it think longer

Pricing shown is typical hosted-inference cost. SWE-bench Pro 11.18% is Scale's public-set leaderboard figure for llama3-1-405b-instruct — third-party, not Meta-reported.

For developers

API model strings

Not researched

Licence

Llama 3.1 Community License Agreement · Restricted · commercial use permitted

Retirement

Retirement date not researched

Lineage

No recorded predecessor

News