Model comparison
DeepSeek LLM 67B vs DeepSeek-V2.5 (Sep 2024)
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 24.9 on the Noometry Index.
Last verified . 10 shared benchmarks.
DeepSeekSummary
- They share 10 benchmarks with published results for both. DeepSeek LLM 67B scores higher in 1 category and DeepSeek-V2.5 (Sep 2024) in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where DeepSeek-V2.5 (Sep 2024) leads 34.8 to 7.0.
Side by side
| DeepSeek LLM 67B | DeepSeek-V2.5 (Sep 2024) | |
|---|---|---|
| Provider | DeepSeek | DeepSeek |
| Noometry Index | 24.9 | 37.6 |
| Released | 2023-11-29 | 2024-09-06 |
| Weights | Open | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 15 | 22 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
DeepSeek LLM 67B: 31.9 (#278), DeepSeek-V2.5 (Sep 2024): 31.7 (#281)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Coding | 1096 | 1309 |
| Aider Polyglot | — | 17.8% |
| BigCodeBench Instruct | — | 48.6% |
| BigCodeBench Complete | — | 53.2% |
| HumanEval+ | — | 83.5% |
| MBPP+ | — | 74.1% |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
DeepSeek LLM 67B: 16.5 (#304), DeepSeek-V2.5 (Sep 2024): 25.6 (#145)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Hard Prompts | 1070 | 1289 |
| Chess Puzzles | 0% | — |
| Epoch Capabilities Index | 110.5 | — |
Math DeepSeek-V2.5 (Sep 2024) leads
DeepSeek LLM 67B: 8.7 (#324), DeepSeek-V2.5 (Sep 2024): 35.9 (#177)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Math | 1108 | 1288 |
| OTIS Mock AIME 2024-2025 | 0.8% | — |
| MATH Level 5 | 6.4% | — |
Knowledge DeepSeek-V2.5 (Sep 2024) leads
DeepSeek LLM 67B: 7.0 (#313), DeepSeek-V2.5 (Sep 2024): 34.8 (#193)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| GPQA Diamond | 24.6% | — |
| LMArena Expert | — | 1266 |
Multilingual DeepSeek-V2.5 (Sep 2024) leads
DeepSeek LLM 67B: 29.4 (#267), DeepSeek-V2.5 (Sep 2024): 42.5 (#193)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Non-English | 1073 | 1273 |
| LMArena Chinese | 1132 | 1318 |
| LMArena French | — | 1289 |
| LMArena German | — | 1258 |
| LMArena Japanese | — | 1228 |
| LMArena Korean | — | 1209 |
| LMArena Russian | — | 1289 |
| LMArena Spanish | — | 1248 |
Instruction Following DeepSeek-V2.5 (Sep 2024) leads
DeepSeek LLM 67B: 55.4 (#277), DeepSeek-V2.5 (Sep 2024): 67.5 (#194)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Instruction Following | 1079 | 1280 |
Long Context DeepSeek-V2.5 (Sep 2024) leads
DeepSeek LLM 67B: 33.1 (#265), DeepSeek-V2.5 (Sep 2024): 39.5 (#174)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Longer Query | 1092 | 1301 |
Writing & Preference DeepSeek-V2.5 (Sep 2024) leads
DeepSeek LLM 67B: 31.6 (#282), DeepSeek-V2.5 (Sep 2024): 49.8 (#187)
| Benchmark | DeepSeek LLM 67B | DeepSeek-V2.5 (Sep 2024) |
|---|---|---|
| LMArena Text | 1105 | 1294 |
| LMArena Creative Writing | 1067 | 1285 |
| LMArena Multi-Turn | 1082 | 1297 |
Frequently asked questions
Is DeepSeek LLM 67B better than DeepSeek-V2.5 (Sep 2024)?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 24.9 on the Noometry Index.
Is DeepSeek LLM 67B or DeepSeek-V2.5 (Sep 2024) better for coding?
They score almost the same on coding (31.9 vs 31.7); test both on your own repository before choosing.
How many benchmarks do DeepSeek LLM 67B and DeepSeek-V2.5 (Sep 2024) share?
10 benchmarks have published results for both models. DeepSeek LLM 67B has 15 scored results on Noometry and DeepSeek-V2.5 (Sep 2024) has 22.