Model comparison
Deepseek Coder v2 vs DeepSeek-V3.1-Terminus
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 42.6 to 35.3 on the Noometry Index.
Last verified . 10 shared benchmarks.
DeepSeekSummary
- They share 10 benchmarks with published results for both. Deepseek Coder v2 scores higher in 0 categories and DeepSeek-V3.1-Terminus in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where DeepSeek-V3.1-Terminus leads 61.3 to 38.4.
Side by side
| Deepseek Coder v2 | DeepSeek-V3.1-Terminus | |
|---|---|---|
| Provider | DeepSeek | DeepSeek |
| Noometry Index | 35.3 | 42.6 |
| Released | 2024-06-17 | 2025-09-22 |
| Weights | Open | Open |
| Context window | — | 164K |
| Max output | — | 66K |
| Input $ / M tokens | — | $0.30 |
| Output $ / M tokens | — | $1 |
| Results tracked | 24 | 16 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.1-Terminus leads
Deepseek Coder v2: 37.5 (#178), DeepSeek-V3.1-Terminus: 41.5 (#114)
| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Coding | 1251 | 1426 |
| SciCode | — | 40.6% |
| BigCodeBench Instruct | 48.2% | — |
| BigCodeBench Complete | 59.7% | — |
| ALE-Bench | — | 745.17 |
| HumanEval+ | 82.3% | — |
| MBPP+ | 75.1% | — |
Reasoning DeepSeek-V3.1-Terminus leads
Deepseek Coder v2: 22.6 (#175), DeepSeek-V3.1-Terminus: 25.7 (#126)
| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Hard Prompts | 1207 | 1427 |
| Kagi LLM Benchmark | — | 57.4% |
| CritPt | — | 1.7% |
| DTBench | — | 81.3% |
| LMCA | — | 28.6% |
| WinoGrande | 83.7% | — |
Math DeepSeek-V3.1-Terminus leads
Deepseek Coder v2: 32.3 (#188), DeepSeek-V3.1-Terminus: 35.7 (#138)
| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Math | 1241 | 1402 |
| GSM8K | 94.5% | — |
Knowledge Not comparable
Deepseek Coder v2: 31.7 (#209), DeepSeek-V3.1-Terminus: —
| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Expert | 1181 | — |
| ARC (AI2) Challenge | 64.3% | — |
Multilingual DeepSeek-V3.1-Terminus leads
Deepseek Coder v2: 36.4 (#235), DeepSeek-V3.1-Terminus: 52.3 (#87)
| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Non-English | 1182 | 1407 |
| LMArena Russian | 1188 | 1434 |
| LMArena Chinese | 1201 | — |
| LMArena French | 1184 | — |
| LMArena German | 1164 | — |
| LMArena Japanese | 1126 | — |
| LMArena Korean | 1104 | — |
| LMArena Spanish | 1153 | — |
Instruction Following DeepSeek-V3.1-Terminus leads
Deepseek Coder v2: 61.9 (#237), DeepSeek-V3.1-Terminus: 74.4 (#96)
| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Instruction Following | 1180 | 1406 |
Long Context DeepSeek-V3.1-Terminus leads
Deepseek Coder v2: 37.0 (#219), DeepSeek-V3.1-Terminus: 43.5 (#88)
| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Longer Query | 1219 | 1426 |
Writing & Preference DeepSeek-V3.1-Terminus leads
Deepseek Coder v2: 38.4 (#249), DeepSeek-V3.1-Terminus: 61.3 (#87)
| Benchmark | Deepseek Coder v2 | DeepSeek-V3.1-Terminus |
|---|---|---|
| LMArena Text | 1191 | 1420 |
| LMArena Creative Writing | 1120 | 1404 |
| LMArena Multi-Turn | 1176 | 1411 |
Frequently asked questions
Is Deepseek Coder v2 better than DeepSeek-V3.1-Terminus?
DeepSeek-V3.1-Terminus is the stronger model overall, scoring 42.6 to 35.3 on the Noometry Index.
Is Deepseek Coder v2 or DeepSeek-V3.1-Terminus better for coding?
DeepSeek-V3.1-Terminus scores higher on coding benchmarks: 41.5 versus 37.5 in the Noometry coding category.
How many benchmarks do Deepseek Coder v2 and DeepSeek-V3.1-Terminus share?
10 benchmarks have published results for both models. Deepseek Coder v2 has 24 scored results on Noometry and DeepSeek-V3.1-Terminus has 16.