Model comparison
Falcon-40B vs Llama 2-34B
Neither Falcon-40B nor Llama 2-34B has enough public benchmark results to be ranked yet; the rows below show what has been published.
Last verified . 10 shared benchmarks.
Technology Innovation Institute
MetaSummary
- They share 10 benchmarks with published results for both.
- Falcon-40B has downloadable open weights; the other is API-only.
Side by side
| Falcon-40B | Llama 2-34B | |
|---|---|---|
| Provider | Technology Innovation Institute | Meta |
| Noometry Index | — | — |
| Released | 2023-03-15 | 2023-07-18 |
| Weights | Open | Proprietary |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 12 | 10 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Reasoning Not comparable
Falcon-40B: —, Llama 2-34B: —
| Benchmark | Falcon-40B | Llama 2-34B |
|---|---|---|
| BIG-Bench Hard | 37.1% | 44.1% |
| Epoch Capabilities Index | 104.45 | 105.2 |
| PIQA | 83% | 81.9% |
| WinoGrande | 76.9% | 76.7% |
| HellaSwag | 85.3% | — |
| LAMBADA | 77.3% | — |
Math Not comparable
Falcon-40B: —, Llama 2-34B: —
| Benchmark | Falcon-40B | Llama 2-34B |
|---|---|---|
| GSM8K | 33.8% | 42.2% |
Knowledge Not comparable
Falcon-40B: —, Llama 2-34B: —
| Benchmark | Falcon-40B | Llama 2-34B |
|---|---|---|
| ARC (AI2) Challenge | 61.9% | 54.5% |
| BoolQ | 83.1% | 83.7% |
| MMLU | 56.9% | 62.6% |
| OpenBookQA | 56.6% | 58.2% |
| TriviaQA | 79.9% | 84.6% |
Frequently asked questions
Is Falcon-40B better than Llama 2-34B?
Neither Falcon-40B nor Llama 2-34B has enough public benchmark results to be ranked yet; the rows below show what has been published.
How many benchmarks do Falcon-40B and Llama 2-34B share?
10 benchmarks have published results for both models. Falcon-40B has 12 scored results on Noometry and Llama 2-34B has 10.