Model comparison
DeepSeek Coder 33B vs Llama 2-7B
Llama 2-7B has enough public results to be ranked (#317); DeepSeek Coder 33B does not yet, so treat this comparison as directional.
Last verified . 5 shared benchmarks.
DeepSeek
MetaSummary
- They share 5 benchmarks with published results for both. DeepSeek Coder 33B scores higher in 1 category and Llama 2-7B in 0 categories; one gap is clear of the uncertainty.
- The widest gap is in coding, where DeepSeek Coder 33B leads 38.0 to 29.2.
Side by side
| DeepSeek Coder 33B | Llama 2-7B | |
|---|---|---|
| Provider | DeepSeek | Meta |
| Noometry Index | 38.9 | 29.1 |
| Released | 2023-11-02 | 2023-07-18 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 9 | 29 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek Coder 33B leads
DeepSeek Coder 33B: 38.0 (#184), Llama 2-7B: 29.2 (#307)
| Benchmark | DeepSeek Coder 33B | Llama 2-7B |
|---|---|---|
| BigCodeBench Instruct | 42% | — |
| LMArena Coding | — | 1002 |
| BigCodeBench Complete | 51.1% | — |
| HumanEval+ | 75% | — |
| MBPP+ | 70.1% | — |
Reasoning Not comparable
DeepSeek Coder 33B: —, Llama 2-7B: 15.7 (#312)
| Benchmark | DeepSeek Coder 33B | Llama 2-7B |
|---|---|---|
| Epoch Capabilities Index | 96.32 | 99.06 |
| WinoGrande | 62% | 69.2% |
| Chess Puzzles | — | 0% |
| LMArena Hard Prompts | — | 1009 |
| BIG-Bench Hard | — | 39.2% |
| HellaSwag | — | 77.2% |
| LAMBADA | — | 73.3% |
| PIQA | — | 78.8% |
Math Not comparable
DeepSeek Coder 33B: —, Llama 2-7B: 30.7 (#233)
| Benchmark | DeepSeek Coder 33B | Llama 2-7B |
|---|---|---|
| GSM8K | 35.4% | 16.7% |
| LMArena Math | — | 1042 |
Knowledge Not comparable
DeepSeek Coder 33B: —, Llama 2-7B: 28.2 (#248)
| Benchmark | DeepSeek Coder 33B | Llama 2-7B |
|---|---|---|
| ARC (AI2) Challenge | 42.2% | 45.9% |
| MMLU | 39.4% | 45.8% |
| LMArena Expert | — | 1036 |
| BoolQ | — | 77.9% |
| OpenBookQA | — | 58.6% |
| TriviaQA | — | 73.7% |
Multimodal Not comparable
DeepSeek Coder 33B: —, Llama 2-7B: —
| Benchmark | DeepSeek Coder 33B | Llama 2-7B |
|---|---|---|
| ScienceQA | — | 43.1% |
Multilingual Not comparable
DeepSeek Coder 33B: —, Llama 2-7B: 23.8 (#293)
| Benchmark | DeepSeek Coder 33B | Llama 2-7B |
|---|---|---|
| LMArena Non-English | — | 973 |
| LMArena Chinese | — | 973 |
| LMArena French | — | 970 |
| LMArena German | — | 978 |
| LMArena Russian | — | 995 |
| LMArena Spanish | — | 1007 |
Instruction Following Not comparable
DeepSeek Coder 33B: —, Llama 2-7B: 50.8 (#298)
| Benchmark | DeepSeek Coder 33B | Llama 2-7B |
|---|---|---|
| LMArena Instruction Following | — | 1006 |
Long Context Not comparable
DeepSeek Coder 33B: —, Llama 2-7B: 30.4 (#287)
| Benchmark | DeepSeek Coder 33B | Llama 2-7B |
|---|---|---|
| LMArena Longer Query | — | 999 |
Writing & Preference Not comparable
DeepSeek Coder 33B: —, Llama 2-7B: 28.0 (#298)
| Benchmark | DeepSeek Coder 33B | Llama 2-7B |
|---|---|---|
| LMArena Text | — | 1053 |
| LMArena Creative Writing | — | 1033 |
| LMArena Multi-Turn | — | 1029 |
Frequently asked questions
Is DeepSeek Coder 33B better than Llama 2-7B?
Llama 2-7B has enough public results to be ranked (#317); DeepSeek Coder 33B does not yet, so treat this comparison as directional.
Is DeepSeek Coder 33B or Llama 2-7B better for coding?
DeepSeek Coder 33B scores higher on coding benchmarks: 38.0 versus 29.2 in the Noometry coding category.
How many benchmarks do DeepSeek Coder 33B and Llama 2-7B share?
5 benchmarks have published results for both models. DeepSeek Coder 33B has 9 scored results on Noometry and Llama 2-7B has 29.