Llama 3.3 70B
Llama 3.3 70B (Meta AI) was released on Dec 6, 2024. In the Epoch AI capability ranking (ECI), it ranks #171 out of 267 models, with 127.34. Against models of the same generation, its best result is on MMLU: 86.3%, #4 of 38.
Where Llama 3.3 70B stands out and where it falls behind, by the numbers
Llama 3.3 70B has published results on 13 tests compiled by Epoch AI. The yardstick here is its generation: its position among the models released between Mar 11, 2024 and Mar 6, 2025 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.
See all 13 tests, one by one
| Test | Score | In its generation | Overall | Test record |
|---|---|---|---|---|
| GPQA Diamond | 47.4% | #19 / 41 | #136 / 173 | 96.4% Gemini 4 Argon |
| AIME 2024 and 2025 (OTIS) | 5.1% | #24 / 35 | #146 / 168 | 100.0% GPT-5.5 |
| Dtbench | 59.5% | #12 / 26 | #117 / 148 | 98.9% Claude Opus 5.5 |
| Lmca | 17.52 | #3 / 13 | #96 / 121 | 68.23 Claude Opus 5.5 |
| CritPt | 0.0% | #2 / 5 (tied with 3) | #91 / 112 | 32.3% GPT-5.6 Sol |
| WeirdML | 14.4% | #15 / 20 | #102 / 112 | 93.6% GPT-6 Astra |
| SciCode | 26.0% | #3 / 4 | #98 / 104 | 66.9% Claude Opus 5.5 |
| MMLU | 86.3% | #4 / 38 | #6 / 92 | 88.1% GPT-4o (May 2024) |
| SimpleBench | 19.9% | #12 / 16 | #70 / 74 | 81.9% Claude Fable 5 |
| MATH level 5 | 41.6% | #24 / 37 | #48 / 73 | 98.1% GPT-5 |
| ForecastBench | 58.60 | #6 / 14 (tied with 1) | #39 / 62 | 62.50 o3 |
| BALROG | 23.0% | #5 / 16 | #21 / 32 | 68.3% GPT-6 Astra |
| LiveBench | 50.16 | #13 / 25 | #17 / 32 | 82.35 Gemini 2.5 Pro (Jun 2025) |
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 3.3 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 |
|---|---|---|---|---|---|
| Qwen2.5-32B | Alibaba | 128.53 | n/a | n/a | Sep 17, 2024 |
| Mistral Small 3.1 | Mistral AI | 127.49 | n/a | n/a | Mar 17, 2025 |
| Llama 3.3 70B | Meta AI | 127.34 | n/a | n/a | Dec 6, 2024 |
| GPT-4 Turbo (Apr 2024) | OpenAI | 127.25 | n/a | n/a | Apr 9, 2024 |
| Claude 3.5 Haiku | Anthropic | 127.17 | 1,286 | n/a | Oct 22, 2024 |
| Mistral Small 3 | Mistral AI | 127.08 | n/a | n/a | Jan 30, 2025 |
| GPT-4o mini | OpenAI | 126.58 | n/a | n/a | Jul 18, 2024 |
Llama 3.3 70B or GPT-4 Turbo (Apr 2024)?
GPT-4 Turbo (Apr 2024) is the model from another lab closest to Llama 3.3 70B in capability. Across the 9 tests both took, this is the score:
| Test | Llama 3.3 70B | GPT-4 Turbo (Apr 2024) |
|---|---|---|
| Lmca | 17.52 | 9.78 |
| AIME 2024 and 2025 (OTIS) | 5.1% | 6.7% |
| SimpleBench | 19.9% | 25.1% |
| WeirdML | 14.4% | 18.0% |
| MATH level 5 | 41.6% | 46.7% |
| MMLU | 86.3% | 81.3% |
| Dtbench | 59.5% | 61.6% |
| GPQA Diamond | 47.4% | 46.6% |
| ForecastBench | 58.60 | 59.40 |
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 4 Maverick | Apr 6, 2025 | 132.20 | n/a | n/a | Open weights (restricted use) |
| Llama 4 Scout | Apr 5, 2025 | 129.65 | n/a | n/a | Open weights (restricted use) |
| Llama 3.3 70B | Dec 6, 2024 | 127.34 | n/a | n/a | Open weights (restricted use) |
| Llama 3.2 90B | Sep 24, 2024 | 125.51 | n/a | n/a | Open weights (restricted use) |
| Llama 3.2 1B | Sep 24, 2024 | 102.04 | n/a | n/a | Open weights (restricted use) |
| Llama 3.1-405B | Jul 23, 2024 | 128.76 | n/a | n/a | Open weights (restricted use) |
| Llama 3.1-70B | Jul 23, 2024 | 125.94 | n/a | n/a | Open weights (restricted use) |
| Llama 3.1-8B | Jul 23, 2024 | 116.61 | n/a | n/a | Open weights (restricted use) |
| Llama 3-70B | Apr 18, 2024 | 122.94 | n/a | n/a | Open weights (restricted use) |
Have you used Llama 3.3 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 3.3 70B
Where does Llama 3.3 70B rank among AI models?
In the ROO3 AI Benchmark, ordered by the Epoch AI ECI, Llama 3.3 70B ranks #171 out of 267 models, with an ECI of 127.34. 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 3.3 70B better than GPT-4 Turbo (Apr 2024)?
Across the 9 tests both models took and Epoch AI compiled, Llama 3.3 70B comes out ahead on 3, behind on 6 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 3.3 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 3.3 70B are: MMLU (#4 of 38); Lmca (#3 of 13); BALROG (#5 of 16). 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