Model comparison
DeepSeek-V3 vs Grok-2 (Dec 2024)
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 33.7 on the Noometry Index.
Last verified . 34 shared benchmarks.
DeepSeekSummary
- They share 34 benchmarks with published results for both. DeepSeek-V3 scores higher in 7 categories and Grok-2 (Dec 2024) in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek-V3 leads 32.1 to 20.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 11.5% for Grok-2 (Dec 2024).
- DeepSeek-V3 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3 | Grok-2 (Dec 2024) | |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 39.5 | 33.7 |
| Released | 2024-12-26 | 2024-08-13 |
| Weights | Open | Proprietary |
| Context window | 164K | — |
| Max output | 164K | — |
| Input $ / M tokens | $0.24 | — |
| Output $ / M tokens | $0.90 | — |
| Results tracked | 60 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 42.3 (#106), Grok-2 (Dec 2024): 33.3 (#258)
| Benchmark | DeepSeek-V3 | Grok-2 (Dec 2024) |
|---|---|---|
| WeirdML | 36.1% | 22.2% |
| LiveBench Coding | 70.9% | 46.4% |
| LMArena Coding | 1368 | 1287 |
| Aider Polyglot | 55.1% | — |
| SciCode | 35.8% | — |
| BigCodeBench Instruct | 50% | — |
| BigCodeBench Complete | 62.2% | — |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, Grok-2 (Dec 2024): —
| Benchmark | DeepSeek-V3 | Grok-2 (Dec 2024) |
|---|---|---|
| METR Time Horizons | 49.6% | — |
Reasoning DeepSeek-V3 leads
DeepSeek-V3: 20.5 (#236), Grok-2 (Dec 2024): 16.9 (#299)
| Benchmark | DeepSeek-V3 | Grok-2 (Dec 2024) |
|---|---|---|
| SimpleBench | 27.2% | 22.7% |
| LiveBench Reasoning | 65.8% | 54.8% |
| LMArena Hard Prompts | 1365 | 1272 |
| DTBench | 64.8% | 65.2% |
| LiveBench Data Analysis | 60.9% | 54.5% |
| Epoch Capabilities Index | 135.94 | 130.48 |
| LiveBench | 66.9% | 54.3% |
| Kagi LLM Benchmark | 52.3% | — |
| CritPt | 0% | — |
| LMCA | 15.5% | — |
| BIG-Bench Hard | 87.5% | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math DeepSeek-V3 leads
DeepSeek-V3: 32.1 (#219), Grok-2 (Dec 2024): 20.8 (#284)
| Benchmark | DeepSeek-V3 | Grok-2 (Dec 2024) |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 11.5% |
| LiveBench Math | 73.5% | 54.9% |
| LMArena Math | 1373 | 1283 |
| MATH Level 5 | 75.5% | 63.5% |
| FrontierMath (Feb 2025 set) | 1.7% | 0.7% |
| Omni-MATH | 40.3% | — |
Knowledge DeepSeek-V3 leads
DeepSeek-V3: 37.5 (#155), Grok-2 (Dec 2024): 29.8 (#233)
| Benchmark | DeepSeek-V3 | Grok-2 (Dec 2024) |
|---|---|---|
| GPQA Diamond | 67.6% | 53.8% |
| Confabulations | 26.1% | 20.1% |
| LMArena Expert | 1351 | 1254 |
| MMLU-Pro | 72.3% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multilingual DeepSeek-V3 leads
DeepSeek-V3: 48.5 (#143), Grok-2 (Dec 2024): 43.1 (#188)
| Benchmark | DeepSeek-V3 | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Non-English | 1358 | 1282 |
| LMArena Chinese | 1391 | 1289 |
| LMArena French | 1385 | 1318 |
| LMArena German | 1374 | 1287 |
| LMArena Japanese | 1333 | 1244 |
| LMArena Korean | 1319 | 1237 |
| LMArena Russian | 1373 | 1286 |
| LMArena Spanish | 1358 | 1281 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 72.8 (#130), Grok-2 (Dec 2024): 66.9 (#202)
| Benchmark | DeepSeek-V3 | Grok-2 (Dec 2024) |
|---|---|---|
| LiveBench Instruction Following | 81.5% | 69.6% |
| LMArena Instruction Following | 1345 | 1270 |
| IFEval | 83.2% | — |
Long Context Grok-2 (Dec 2024) leads
DeepSeek-V3: 34.0 (#253), Grok-2 (Dec 2024): 38.8 (#190)
| Benchmark | DeepSeek-V3 | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Longer Query | 1352 | 1276 |
| Fiction.LiveBench | 50% | — |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.4 (#130), Grok-2 (Dec 2024): 48.6 (#198)
| Benchmark | DeepSeek-V3 | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Text | 1375 | 1305 |
| LMArena Creative Writing | 1364 | 1284 |
| Short-Story Creative Writing | 77% | 63.6% |
| LMArena Multi-Turn | 1389 | 1290 |
| LiveBench Language | 49.1% | 45.6% |
| EQ-Bench Creative Writing | 1472 | — |
| WildBench | 83% | — |
Frequently asked questions
Is DeepSeek-V3 better than Grok-2 (Dec 2024)?
DeepSeek-V3 is the stronger model overall, scoring 39.5 to 33.7 on the Noometry Index.
Is DeepSeek-V3 or Grok-2 (Dec 2024) better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 42.3 versus 33.3 in the Noometry coding category.
How many benchmarks do DeepSeek-V3 and Grok-2 (Dec 2024) share?
34 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and Grok-2 (Dec 2024) has 34.