Model comparison
Codellama 70b Instruct vs Deepseek Coder v2
Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 33.7 on the Noometry Index.
Last verified . 7 shared benchmarks.
Meta
DeepSeekSummary
- They share 7 benchmarks with published results for both. Codellama 70b Instruct scores higher in 0 categories and Deepseek Coder v2 in 5 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in multilingual, where Deepseek Coder v2 leads 36.3 to 24.8.
- The biggest single-benchmark swing is BigCodeBench Complete: 49.6% for Codellama 70b Instruct and 59.7% for Deepseek Coder v2.
Side by side
| Codellama 70b Instruct | Deepseek Coder v2 | |
|---|---|---|
| Provider | Meta | DeepSeek |
| Noometry Index | 33.7 | 35.9 |
| Released | — | 2024-06-17 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 7 | 24 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Codellama 70b Instruct: 37.6 (#193), Deepseek Coder v2: 38.1 (#183)
| Benchmark | Codellama 70b Instruct | Deepseek Coder v2 |
|---|---|---|
| BigCodeBench Instruct | 40.7% | 48.2% |
| BigCodeBench Complete | 49.6% | 59.7% |
| HumanEval+ | 65.9% | 82.3% |
| LMArena Coding | — | 1251 |
| MBPP+ | — | 75.1% |
Reasoning Deepseek Coder v2 leads
Codellama 70b Instruct: 20.1 (#242), Deepseek Coder v2: 23.6 (#176)
| Benchmark | Codellama 70b Instruct | Deepseek Coder v2 |
|---|---|---|
| LMArena Hard Prompts | 1052 | 1207 |
| WinoGrande | — | 83.7% |
Math Not comparable
Codellama 70b Instruct: —, Deepseek Coder v2: 34.9 (#190)
| Benchmark | Codellama 70b Instruct | Deepseek Coder v2 |
|---|---|---|
| LMArena Math | — | 1241 |
| GSM8K | — | 94.5% |
Knowledge Not comparable
Codellama 70b Instruct: —, Deepseek Coder v2: 32.3 (#212)
| Benchmark | Codellama 70b Instruct | Deepseek Coder v2 |
|---|---|---|
| LMArena Expert | — | 1181 |
| ARC (AI2) Challenge | — | 64.3% |
Multilingual Deepseek Coder v2 leads
Codellama 70b Instruct: 24.8 (#288), Deepseek Coder v2: 36.3 (#240)
| Benchmark | Codellama 70b Instruct | Deepseek Coder v2 |
|---|---|---|
| LMArena Non-English | 992 | 1182 |
| LMArena Chinese | — | 1201 |
| LMArena French | — | 1185 |
| LMArena German | — | 1164 |
| LMArena Japanese | — | 1126 |
| LMArena Korean | — | 1104 |
| LMArena Russian | — | 1188 |
| LMArena Spanish | — | 1153 |
Instruction Following Deepseek Coder v2 leads
Codellama 70b Instruct: 51.9 (#293), Deepseek Coder v2: 61.7 (#242)
| Benchmark | Codellama 70b Instruct | Deepseek Coder v2 |
|---|---|---|
| LMArena Instruction Following | 1024 | 1180 |
Long Context Not comparable
Codellama 70b Instruct: —, Deepseek Coder v2: 37.0 (#224)
| Benchmark | Codellama 70b Instruct | Deepseek Coder v2 |
|---|---|---|
| LMArena Longer Query | — | 1219 |
Writing & Preference Deepseek Coder v2 leads
Codellama 70b Instruct: 33.4 (#277), Deepseek Coder v2: 38.2 (#253)
| Benchmark | Codellama 70b Instruct | Deepseek Coder v2 |
|---|---|---|
| LMArena Text | 1057 | 1191 |
| LMArena Creative Writing | — | 1120 |
| LMArena Multi-Turn | — | 1177 |
Frequently asked questions
Is Codellama 70b Instruct better than Deepseek Coder v2?
Deepseek Coder v2 is the stronger model overall, scoring 35.9 to 33.7 on the Noometry Index.
Is Codellama 70b Instruct or Deepseek Coder v2 better for coding?
They score almost the same on coding (37.6 vs 38.1); test both on your own repository before choosing.
How many benchmarks do Codellama 70b Instruct and Deepseek Coder v2 share?
7 benchmarks have published results for both models. Codellama 70b Instruct has 7 scored results on Noometry and Deepseek Coder v2 has 24.