Model comparison
Deepseek Coder v2 vs Llama 3.1 Tulu 3 8b
Deepseek Coder v2 and Llama 3.1 Tulu 3 8b score almost the same on the Noometry Index (35.9 vs 35.7), so choose on price, context window or the category you care about most.
Last verified . 11 shared benchmarks.
DeepSeek
Allen Institute for AI (Ai2)Summary
- They share 11 benchmarks with published results for both. Deepseek Coder v2 scores higher in 6 categories and Llama 3.1 Tulu 3 8b in 1 category; 4 gaps are clear of the uncertainty.
- The widest gap is in coding, where Deepseek Coder v2 leads 38.1 to 34.4.
Side by side
| Deepseek Coder v2 | Llama 3.1 Tulu 3 8b | |
|---|---|---|
| Provider | DeepSeek | Allen Institute for AI (Ai2) |
| Noometry Index | 35.9 | 35.7 |
| Released | 2024-06-17 | — |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 24 | 11 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Deepseek Coder v2 leads
Deepseek Coder v2: 38.1 (#183), Llama 3.1 Tulu 3 8b: 34.4 (#235)
| Benchmark | Deepseek Coder v2 | Llama 3.1 Tulu 3 8b |
|---|---|---|
| LMArena Coding | 1251 | 1183 |
| BigCodeBench Instruct | 48.2% | — |
| BigCodeBench Complete | 59.7% | — |
| HumanEval+ | 82.3% | — |
| MBPP+ | 75.1% | — |
Reasoning Too close to call
Deepseek Coder v2: 23.6 (#176), Llama 3.1 Tulu 3 8b: 22.8 (#188)
| Benchmark | Deepseek Coder v2 | Llama 3.1 Tulu 3 8b |
|---|---|---|
| LMArena Hard Prompts | 1207 | 1174 |
| WinoGrande | 83.7% | — |
Math Deepseek Coder v2 leads
Deepseek Coder v2: 34.9 (#190), Llama 3.1 Tulu 3 8b: 33.9 (#198)
| Benchmark | Deepseek Coder v2 | Llama 3.1 Tulu 3 8b |
|---|---|---|
| LMArena Math | 1241 | 1195 |
| GSM8K | 94.5% | — |
Knowledge Not comparable
Deepseek Coder v2: 32.3 (#212), Llama 3.1 Tulu 3 8b: —
| Benchmark | Deepseek Coder v2 | Llama 3.1 Tulu 3 8b |
|---|---|---|
| LMArena Expert | 1181 | — |
| ARC (AI2) Challenge | 64.3% | — |
Multilingual Too close to call
Deepseek Coder v2: 36.3 (#240), Llama 3.1 Tulu 3 8b: 35.4 (#246)
| Benchmark | Deepseek Coder v2 | Llama 3.1 Tulu 3 8b |
|---|---|---|
| LMArena Non-English | 1182 | 1169 |
| LMArena Chinese | 1201 | 1176 |
| LMArena Russian | 1188 | 1193 |
| LMArena French | 1185 | — |
| LMArena German | 1164 | — |
| LMArena Japanese | 1126 | — |
| LMArena Korean | 1104 | — |
| LMArena Spanish | 1153 | — |
Instruction Following Too close to call
Deepseek Coder v2: 61.7 (#242), Llama 3.1 Tulu 3 8b: 61.3 (#246)
| Benchmark | Deepseek Coder v2 | Llama 3.1 Tulu 3 8b |
|---|---|---|
| LMArena Instruction Following | 1180 | 1174 |
Long Context Deepseek Coder v2 leads
Deepseek Coder v2: 37.0 (#224), Llama 3.1 Tulu 3 8b: 35.8 (#239)
| Benchmark | Deepseek Coder v2 | Llama 3.1 Tulu 3 8b |
|---|---|---|
| LMArena Longer Query | 1219 | 1181 |
Writing & Preference Llama 3.1 Tulu 3 8b leads
Deepseek Coder v2: 38.2 (#253), Llama 3.1 Tulu 3 8b: 39.7 (#245)
| Benchmark | Deepseek Coder v2 | Llama 3.1 Tulu 3 8b |
|---|---|---|
| LMArena Text | 1191 | 1193 |
| LMArena Creative Writing | 1120 | 1182 |
| LMArena Multi-Turn | 1177 | 1154 |
Frequently asked questions
Is Deepseek Coder v2 better than Llama 3.1 Tulu 3 8b?
Deepseek Coder v2 and Llama 3.1 Tulu 3 8b score almost the same on the Noometry Index (35.9 vs 35.7), so choose on price, context window or the category you care about most.
Is Deepseek Coder v2 or Llama 3.1 Tulu 3 8b better for coding?
Deepseek Coder v2 scores higher on coding benchmarks: 38.1 versus 34.4 in the Noometry coding category.
How many benchmarks do Deepseek Coder v2 and Llama 3.1 Tulu 3 8b share?
11 benchmarks have published results for both models. Deepseek Coder v2 has 24 scored results on Noometry and Llama 3.1 Tulu 3 8b has 11.