Llama 3

Announced Apr 18, 2024

Downloadable models become more useful

What could it do?

Meta released two sizes that developers could build on and customize, with better reasoning and coding than before.

What changed?

Improved instruction following and coding over Llama 2.

WHY IT MATTERED

Model access

Gave builders another capable model family with downloadable weights.

Where it fell short

License terms and hosting requirements still apply.

Research draft

Capabilities describe the developer’s announcement. This entry has not completed source review.

Read original source

Benchmark results

EPOCH AI CAPABILITY ESTIMATE

122.9index points

Tested variant: Llama 3-70BSource interval: 116.6 to 125.8Variant date in source: Apr 18, 2024

A benchmark estimate, not a percentage or capability multiplier. Reasoning settings are not specified in this source table. Historical estimates can change in later snapshots.

Epoch AI methodology ↗Download the source snapshotChecked Oct 7, 2026 · CC BY 4.0

PUBLISHED BENCHMARK RESULT

1277rating points

Tested: llama-3-70b-instructText Arena Overall · October 8, 2026Reported interval: 1273 to 1281156,876 votes

One explicitly named variant per release. Scores come from the same Overall snapshot; preliminary entries and reported intervals are preserved. These are current ratings of earlier variants, not their launch day ratings.

Source: Text Arena ↗Download selected resultsChecked Oct 8, 2026

Understand the benchmarks Find this release on the chart