Model comparison
DeepSeek LLM 67B vs DeepSeek-V3.1
DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 24.9 on the Noometry Index.
Last verified . 11 shared benchmarks.
DeepSeekSummary
- They share 11 benchmarks with published results for both. DeepSeek LLM 67B scores higher in 0 categories and DeepSeek-V3.1 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V3.1 leads 43.7 to 7.0.
Side by side
| DeepSeek LLM 67B | DeepSeek-V3.1 | |
|---|---|---|
| Provider | DeepSeek | DeepSeek |
| Noometry Index | 24.9 | 42.8 |
| Released | 2023-11-29 | 2025-08-21 |
| Weights | Open | Open |
| Context window | — | 164K |
| Max output | — | 8K |
| Input $ / M tokens | — | $0.25 |
| Output $ / M tokens | — | $0.95 |
| Results tracked | 15 | 27 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.1 leads
DeepSeek LLM 67B: 31.9 (#278), DeepSeek-V3.1: 40.3 (#144)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.1 |
|---|---|---|
| LMArena Coding | 1096 | 1417 |
| WeirdML | — | 38.4% |
Reasoning DeepSeek-V3.1 leads
DeepSeek LLM 67B: 16.5 (#304), DeepSeek-V3.1: 27.9 (#110)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.1 |
|---|---|---|
| LMArena Hard Prompts | 1070 | 1417 |
| Epoch Capabilities Index | 110.5 | 139.92 |
| SimpleBench | — | 40% |
| Kagi LLM Benchmark | — | 53.2% |
| Chess Puzzles | 0% | — |
| DTBench | — | 82.7% |
| LMCA | — | 24.3% |
| ForecastBench | — | 58 |
Math DeepSeek-V3.1 leads
DeepSeek LLM 67B: 8.7 (#324), DeepSeek-V3.1: 38.9 (#122)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.1 |
|---|---|---|
| LMArena Math | 1108 | 1420 |
| OTIS Mock AIME 2024-2025 | 0.8% | — |
| MATH Level 5 | 6.4% | — |
Knowledge DeepSeek-V3.1 leads
DeepSeek LLM 67B: 7.0 (#313), DeepSeek-V3.1: 43.7 (#90)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.1 |
|---|---|---|
| GPQA Diamond | 24.6% | — |
| Vectara Hallucination Rate | — | 5.5% |
| LMArena Expert | — | 1405 |
Multilingual DeepSeek-V3.1 leads
DeepSeek LLM 67B: 29.4 (#267), DeepSeek-V3.1: 51.6 (#106)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.1 |
|---|---|---|
| LMArena Non-English | 1073 | 1400 |
| LMArena Chinese | 1132 | 1469 |
| LMArena French | — | 1447 |
| LMArena German | — | 1411 |
| LMArena Japanese | — | 1378 |
| LMArena Korean | — | 1337 |
| LMArena Russian | — | 1405 |
| LMArena Spanish | — | 1431 |
Instruction Following DeepSeek-V3.1 leads
DeepSeek LLM 67B: 55.4 (#277), DeepSeek-V3.1: 73.9 (#110)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.1 |
|---|---|---|
| LMArena Instruction Following | 1079 | 1400 |
Long Context DeepSeek-V3.1 leads
DeepSeek LLM 67B: 33.1 (#265), DeepSeek-V3.1: 36.3 (#232)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.1 |
|---|---|---|
| LMArena Longer Query | 1092 | 1422 |
| Fiction.LiveBench | — | 52.8% |
Writing & Preference DeepSeek-V3.1 leads
DeepSeek LLM 67B: 31.6 (#282), DeepSeek-V3.1: 60.3 (#98)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V3.1 |
|---|---|---|
| LMArena Text | 1105 | 1420 |
| LMArena Creative Writing | 1067 | 1401 |
| LMArena Multi-Turn | 1082 | 1408 |
| EQ-Bench Creative Writing | — | 1436 |
Frequently asked questions
Is DeepSeek LLM 67B better than DeepSeek-V3.1?
DeepSeek-V3.1 is the stronger model overall, scoring 42.8 to 24.9 on the Noometry Index.
Is DeepSeek LLM 67B or DeepSeek-V3.1 better for coding?
DeepSeek-V3.1 scores higher on coding benchmarks: 40.3 versus 31.9 in the Noometry coding category.
How many benchmarks do DeepSeek LLM 67B and DeepSeek-V3.1 share?
11 benchmarks have published results for both models. DeepSeek LLM 67B has 15 scored results on Noometry and DeepSeek-V3.1 has 27.