Meta · open weights on Arena · first on Arena 30 July 2024

llama-3.1-405b-instruct

179 days for open weights to match gpt-4-0125-preview

This open-weights model matched the score of gpt-4-0125-preview in Text 179 days after gpt-4-0125-preview took first place. It also matched gpt-4-1106-preview in Text, claude-3-opus-20240229 in Text and gpt-4-turbo-2024-04-09 in Text.

Embed

Who held first place

<iframe src="https://jesseahawkins.com/frontier/embed/" title="Frontier: who held first place on each Arena leaderboard, and for how long" width="100%" height="560" style="border:0;max-width:100%" loading="lazy"></iframe>

Credits LMArena.

The open-weights catch-up

<iframe src="https://jesseahawkins.com/frontier/embed/catch-up/" title="Frontier: the days until open weights matched each closed leader" width="100%" height="900" style="border:0;max-width:100%" loading="lazy"></iframe>

Credits LMArena.

The saturation clock

<iframe src="https://jesseahawkins.com/frontier/embed/clock/" title="Frontier: the days from each benchmark’s release to a model at 90%" width="100%" height="640" style="border:0;max-width:100%" loading="lazy"></iframe>

Credits Epoch AI.

Each frame links back to its page. Preview it.

Cite

This page

Jesse Hawkins, “llama-3.1-405b-instruct · Frontier”, jesseahawkins.com, data retrieved 30 Oct 2026. https://jesseahawkins.com/frontier/llama-3-1-405b-instruct/. Data: LMArena, Arena leaderboard dataset, revision 4491b06, CC BY 4.0; Epoch AI, Capabilities & benchmarking, CC BY 4.0.

BibTeX

@misc{hawkins2026frontierllama31405binstruct,
  author = {Hawkins, Jesse},
  title = {{llama-3.1-405b-instruct · Frontier}},
  year = {2026},
  howpublished = {\url{https://jesseahawkins.com/frontier/llama-3-1-405b-instruct/}},
  note = {Data: LMArena, Arena leaderboard dataset, revision 4491b06, CC BY 4.0; Epoch AI, Capabilities \& benchmarking, CC BY 4.0, retrieved 2026-10-30}
}

Arena · LMArena

LMArena, "Arena leaderboard dataset", Hugging Face, revision 4491b06. https://huggingface.co/datasets/lmarena-ai/leaderboard-dataset. CC BY 4.0.

Benchmarks · Epoch AI

Epoch AI, 'Capabilities & benchmarking'. Published online at epoch.ai. Retrieved from 'https://epoch.ai/benchmarks' [online resource]. CC BY 4.0.

Download

Post on X

Text: best place 5

last on the board 4 September 2024

1st10th

On the board: 30 July 2024 to 4 September 2024
Score: 1266 (1262–1271) on 4 September 2024 · 27,397 votes
Matched: gpt-4-0125-preview on 30 July 2024, 179 days after it took first place
gpt-4-1106-preview on 30 July 2024, 166 days after it took first place
claude-3-opus-20240229 on 30 July 2024, 126 days after it took first place
gpt-4-turbo-2024-04-09 on 30 July 2024, 110 days after it took first place

Epoch AI: Llama 3.1-405B

released 23 July 2024 · Open weights (restricted use)

GPQA diamond 50.9%
MATH level 5 49.8%
OTIS Mock AIME 2024-2025 9.7%

Sources

Arena scores come from people voting between two anonymous answers. First place is the highest score; the top two are often within each other’s margin of error. LMArena, Arena leaderboard dataset, revision 4491b06, retrieved 30 October 2026; Epoch AI, Capabilities & benchmarking, retrieved 30 October 2026. Both CC BY 4.0.