Llama 2-70B
Llama 2-70B (Meta AI) was released on Jul 18, 2023. In the Epoch AI capability ranking (ECI), it ranks #212 out of 267 models, with 113.84. Against models of the same generation, its best result is on TriviaQA: 87.6%, #1 of 19.
Where Llama 2-70B stands out and where it falls behind, by the numbers
Llama 2-70B has published results on 17 tests compiled by Epoch AI. The yardstick here is its generation: its position among the models released between Oct 21, 2022 and Oct 16, 2023 that took the same test. Against the whole archive, which includes 2023 models, a recent release lands near the top of almost everything, and that says nothing.
Where it stands out
Tests where it sits in the top third of its own generation.
Where it falls behind
Tests where it sits in the bottom half of its own generation.
It is not in the bottom half of its generation on any test with 6 or more models compared.
See all 17 tests, one by one
| Test | Score | In its generation | Overall | Test record |
|---|---|---|---|---|
| GPQA Diamond | 26.3% | #3 / 3 | #166 / 173 | 96.4% Gemini 4 Argon |
| AIME 2024 and 2025 (OTIS) | 0.0% | #3 / 3 | #168 / 168 | 100.0% GPT-5.5 |
| Dtbench | 41.7% | #5 / 5 | #148 / 148 | 98.9% Claude Opus 5.5 |
| MMLU | 69.9% | #5 / 31 | #41 / 92 | 88.1% GPT-4o (May 2024) |
| MATH level 5 | 3.3% | #3 / 3 | #73 / 73 | 98.1% GPT-5 |
| ARC (AI2) | 78.3% | #6 / 34 | #16 / 65 | 95.3% DeepSeek-V3 (Mar 2025) |
| ForecastBench | 51.40 | #2 / 2 | #61 / 62 | 62.50 o3 |
| WinoGrande | 80.2% | #4 / 26 | #14 / 61 | 89.2% Llama 3.1-405B |
| GSM8K | 69.6% | #3 / 26 (tied with 1) | #14 / 54 | 92.0% GPT-4 (Mar 2023) |
| HellaSwag | 85.3% | #4 / 30 | #8 / 51 | 95.3% GPT-4 (Mar 2023) |
| PIQA | 82.8% | #5 / 29 (tied with 1) | #13 / 43 | 88.7% GPT-4o mini |
| BoolQ | 88.6% | #4 / 33 (tied with 1) | #5 / 42 | 90.9% PaLM 2-L |
| BIG-Bench Hard | 64.9% | #3 / 24 | #12 / 41 | 89.2% Gemini 1.5 Pro (Sept 2024) |
| OpenBookQA | 60.2% | #3 / 24 (tied with 1) | #11 / 32 | 88.0% phi-3-mini 3.8B |
| TriviaQA | 87.6% | #1 / 19 | #1 / 32 | 87.6% Llama 2-70B |
| LAMBADA | 78.9% | #2 / 19 | #2 / 20 | 79.8% Falcon-180B |
| Common Sense Qa 2 | 50.0% | #1 / 1 | #2 / 2 | 57.0% GPT-3.5 Turbo (Nov 2023) |
For each test, the model's best effort level counts, as in the ranking. Source of the scores: Epoch AI, under CC BY 4.0.
What to compare Llama 2-70B with
The models from other labs closest in capability and released in the same period, at most two per lab. Click any of them to see its page.
| Model | Lab | ECI | ELO | LiveBench | Release |
|---|---|---|---|---|---|
| PaLM 2-L | 115.07 | n/a | n/a | May 17, 2023 | |
| Llama 2-70B | Meta AI | 113.84 | n/a | n/a | Jul 18, 2023 |
| Qwen-14B | Alibaba | 113.09 | n/a | n/a | Sep 24, 2023 |
| Mistral 7B v0.1 | Mistral AI | 112.27 | n/a | n/a | Sep 27, 2023 |
| Falcon-180B | Technology Innovation Institute | 112.19 | n/a | n/a | Sep 6, 2023 |
| internlm-20b | 112.15 | n/a | n/a | Sep 18, 2023 | |
| Gemma 7B | Google DeepMind | 112.05 | n/a | n/a | Feb 21, 2024 |
Llama 2-70B or Qwen-14B?
Qwen-14B is the model from another lab closest to Llama 2-70B in capability. Across the 7 tests both took, this is the score:
| Test | Llama 2-70B | Qwen-14B |
|---|---|---|
| BIG-Bench Hard | 64.9% | 53.4% |
| GSM8K | 69.6% | 61.3% |
| LAMBADA | 78.9% | 71.1% |
| ARC (AI2) | 78.3% | 84.4% |
| MMLU | 69.9% | 66.3% |
| PIQA | 82.8% | 79.9% |
| BoolQ | 88.6% | 86.2% |
Unweighted score: each test counts one point, whether math or coding. The dot marks who did better on each row.
Other models from Meta AI
The models from the same lab closest in time, newest first.
| Model | Release | ECI | ELO | LiveBench | Access |
|---|---|---|---|---|---|
| Llama 3-70B | Apr 18, 2024 | 122.94 | n/a | n/a | Open weights (restricted use) |
| Llama 2-70B | Jul 18, 2023 | 113.84 | n/a | n/a | Open weights (restricted use) |
| Llama 2-13B | Jul 18, 2023 | 106.25 | n/a | n/a | Open weights (restricted use) |
| Llama 2-34B | Jul 18, 2023 | 105.29 | n/a | n/a | Unreleased |
| Llama 2-7B | Jul 18, 2023 | 99.18 | n/a | n/a | Open weights (restricted use) |
| LLaMA-65B | Feb 24, 2023 | 110.23 | n/a | n/a | Open weights (non-commercial) |
| LLaMA-33B | Feb 24, 2023 | 107.49 | n/a | n/a | Open weights (non-commercial) |
| LLaMA-13B | Feb 24, 2023 | 100.69 | 846 | n/a | Open weights (non-commercial) |
| LLaMA-7B | Feb 24, 2023 | 96.82 | n/a | n/a | Open weights (non-commercial) |
Have you used Llama 2-70B? Rate it
Users' opinion, not measurement: the rating is kept apart from the ranking and does not move anyone's position. One vote per browser, no sign-up, and you can change your rating later.
What people ask about Llama 2-70B
Where does Llama 2-70B rank among AI models?
In the ROO3 AI Benchmark, ordered by the Epoch AI ECI, Llama 2-70B ranks #212 out of 267 models, with an ECI of 113.84. The ECI is a composite index that combines several hard benchmarks into a single scale. The position changes when a new model comes out, and the page is refreshed six times a day.
Is Llama 2-70B better than Qwen-14B?
Across the 7 tests both models took and Epoch AI compiled, Llama 2-70B comes out ahead on 6, behind on 1 and tied on 0. It is a test-by-test count with no weighting: a win in math counts the same as a win in coding. To choose between the two, look at the tests that resemble your own use.
What is Llama 2-70B best at?
Against models of the same generation (released up to 9 months before and up to 3 months after it), the best relative results of Llama 2-70B are: TriviaQA (#1 of 19); LAMBADA (#2 of 19); BIG-Bench Hard (#3 of 24). The yardstick is its position among the models of that period that took the same test.
Where do the numbers on this page come from?
From the same three sources as the ranking, all openly licensed: Epoch AI (ECI and individual tests, CC BY 4.0), Arena (human-preference ELO, CC BY 4.0) and LiveBench (overall average, CC BY-SA 4.0). ROO3 does not measure any model. When the page shows a figure published by the maker itself, it appears in the written analysis, labeled as the maker's, and never enters the ranking.
Where each number comes from
Epoch AI, "AI Benchmarking Hub". Published at epoch.ai. Data under the CC BY 4.0 licence, reordered by roo3.co.
The Arena (LMArena) leaderboard, from the public lmarena-ai/leaderboard-dataset on Hugging Face, under the CC BY 4.0 licence. We are not affiliated with the Arena.
LiveBench, by the LiveBench team, published at livebench.ai under the CC BY-SA 4.0 license. The overall average is recomputed by roo3.co with the same formula as the website (average of the categories), and those numbers stay under CC BY-SA 4.0. We are not affiliated with LiveBench.
← See the full ranking