Llama 2

Announced Jul 18, 2023

More accessible model weights

What could it do?

Meta opened it to startups, researchers, and businesses so they could build their own products on it.

What changed?

Broadened access to capable pretrained and chat models.

WHY IT MATTERED

Model access

Helped developers experiment outside closed APIs, subject to its license.

Where it fell short

Requires hosting resources and careful deployment.

Research draft

Capabilities describe the developer’s announcement. This entry has not completed source review. This entry describes the model family. Capabilities vary by model size.

Read original source

Benchmark results

EPOCH AI CAPABILITY ESTIMATE

113.8index points

Tested variant: Llama 2-70BSource interval: 105.7 to 117.4Variant date in source: Jul 18, 2023

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

1171rating points

Tested: llama-2-70b-chatText Arena Overall · October 8, 2026Reported interval: 1166 to 117638,492 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