Model comparison
DeepSeek-V2.5 (Sep 2024) vs Grok-2 (Dec 2024)
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 33.7 on the Noometry Index.
Last verified . 17 shared benchmarks.
DeepSeekSummary
- They share 17 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 6 categories and Grok-2 (Dec 2024) in 2 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-V2.5 (Sep 2024) leads 35.9 to 20.8.
- DeepSeek-V2.5 (Sep 2024) has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V2.5 (Sep 2024) | Grok-2 (Dec 2024) | |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 37.6 | 33.7 |
| Released | 2024-09-06 | 2024-08-13 |
| Weights | Open | Proprietary |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 22 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok-2 (Dec 2024) leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), Grok-2 (Dec 2024): 33.3 (#258)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Coding | 1309 | 1287 |
| Aider Polyglot | 17.8% | — |
| WeirdML | — | 22.2% |
| BigCodeBench Instruct | 48.6% | — |
| LiveBench Coding | — | 46.4% |
| BigCodeBench Complete | 53.2% | — |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Reasoning DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), Grok-2 (Dec 2024): 16.9 (#299)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1272 |
| SimpleBench | — | 22.7% |
| LiveBench Reasoning | — | 54.8% |
| DTBench | — | 65.2% |
| LiveBench Data Analysis | — | 54.5% |
| Epoch Capabilities Index | — | 130.48 |
| LiveBench | — | 54.3% |
Math DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), Grok-2 (Dec 2024): 20.8 (#284)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Math | 1288 | 1283 |
| OTIS Mock AIME 2024-2025 | — | 11.5% |
| LiveBench Math | — | 54.9% |
| MATH Level 5 | — | 63.5% |
| FrontierMath (Feb 2025 set) | — | 0.7% |
Knowledge DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), Grok-2 (Dec 2024): 29.8 (#233)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Expert | 1266 | 1254 |
| GPQA Diamond | — | 53.8% |
| Confabulations | — | 20.1% |
Multilingual Too close to call
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), Grok-2 (Dec 2024): 43.1 (#188)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Non-English | 1273 | 1282 |
| LMArena Chinese | 1318 | 1289 |
| LMArena French | 1289 | 1318 |
| LMArena German | 1258 | 1287 |
| LMArena Japanese | 1228 | 1244 |
| LMArena Korean | 1209 | 1237 |
| LMArena Russian | 1289 | 1286 |
| LMArena Spanish | 1248 | 1281 |
Instruction Following Too close to call
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), Grok-2 (Dec 2024): 66.9 (#202)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Instruction Following | 1280 | 1270 |
| LiveBench Instruction Following | — | 69.6% |
Long Context Too close to call
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), Grok-2 (Dec 2024): 38.8 (#190)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Longer Query | 1301 | 1276 |
Writing & Preference DeepSeek-V2.5 (Sep 2024) leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), Grok-2 (Dec 2024): 48.6 (#198)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Text | 1294 | 1305 |
| LMArena Creative Writing | 1285 | 1284 |
| LMArena Multi-Turn | 1297 | 1290 |
| Short-Story Creative Writing | — | 63.6% |
| LiveBench Language | — | 45.6% |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than Grok-2 (Dec 2024)?
DeepSeek-V2.5 (Sep 2024) is the stronger model overall, scoring 37.6 to 33.7 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or Grok-2 (Dec 2024) better for coding?
Grok-2 (Dec 2024) scores higher on coding benchmarks: 33.3 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and Grok-2 (Dec 2024) share?
17 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and Grok-2 (Dec 2024) has 34.