Model comparison
DeepSeek V4 Pro vs Grok-2 (Dec 2024)
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 33.7 on the Noometry Index.
Last verified . 22 shared benchmarks.
DeepSeekSummary
- They share 22 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 8 categories and Grok-2 (Dec 2024) in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where DeepSeek V4 Pro leads 64.8 to 20.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 98.6% for DeepSeek V4 Pro and 11.5% for Grok-2 (Dec 2024).
- DeepSeek V4 Pro has downloadable open weights; the other is API-only.
Side by side
| DeepSeek V4 Pro | Grok-2 (Dec 2024) | |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 54.3 | 33.7 |
| Released | 2026-04-24 | 2024-08-13 |
| Weights | Open | Proprietary |
| Context window | 1M | — |
| Max output | 393K | — |
| Input $ / M tokens | $0.66 | — |
| Output $ / M tokens | $1.98 | — |
| Results tracked | 48 | 34 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Pro leads
DeepSeek V4 Pro: 52.4 (#34), Grok-2 (Dec 2024): 33.3 (#258)
| Benchmark | DeepSeek V4 Pro | Grok-2 (Dec 2024) |
|---|---|---|
| WeirdML | 66.2% | 22.2% |
| LMArena Coding | 1470 | 1287 |
| SWE-bench Verified | 77.6% | — |
| FrontierCode | 28.6% | — |
| LMArena WebDev | 1582 | — |
| SciCode | 51% | — |
| LiveBench Coding | — | 46.4% |
| ALE-Bench | 1,403 | — |
Agentic & Tool Use Not comparable
DeepSeek V4 Pro: 32.8 (#58), Grok-2 (Dec 2024): —
| Benchmark | DeepSeek V4 Pro | Grok-2 (Dec 2024) |
|---|---|---|
| APEX-Agents | 47.3% | — |
| Vending-Bench 2 | 3,285 | — |
Reasoning DeepSeek V4 Pro leads
DeepSeek V4 Pro: 56.5 (#24), Grok-2 (Dec 2024): 16.9 (#299)
| Benchmark | DeepSeek V4 Pro | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Hard Prompts | 1461 | 1272 |
| DTBench | 93.9% | 65.2% |
| Epoch Capabilities Index | 155.31 | 130.48 |
| ARC-AGI-2 | 61.3% | — |
| SimpleBench | — | 22.7% |
| Kagi LLM Benchmark | 53.5% | — |
| NYT Connections (extended) | 91.3% | — |
| ARC-AGI-1 | 90.5% | — |
| CritPt | 18% | — |
| Chess Puzzles | 47% | — |
| LiveBench Reasoning | — | 54.8% |
| Mystery Game Puzzles | 43% | — |
| LiveBench Data Analysis | — | 54.5% |
| LMCA | 45.5% | — |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |
| LiveBench | — | 54.3% |
Math DeepSeek V4 Pro leads
DeepSeek V4 Pro: 64.8 (#30), Grok-2 (Dec 2024): 20.8 (#284)
| Benchmark | DeepSeek V4 Pro | Grok-2 (Dec 2024) |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 98.6% | 11.5% |
| LMArena Math | 1455 | 1283 |
| FrontierMath (Tiers 1-3) | 64.6% | — |
| FrontierMath Tier 4 | 26.8% | — |
| MathArena Final-Answer Competitions | 76.6% | — |
| ProofBench | 50% | — |
| LiveBench Math | — | 54.9% |
| MATH Level 5 | — | 63.5% |
| FrontierMath (Feb 2025 set) | — | 0.7% |
Knowledge DeepSeek V4 Pro leads
DeepSeek V4 Pro: 59.5 (#31), Grok-2 (Dec 2024): 29.8 (#233)
| Benchmark | DeepSeek V4 Pro | Grok-2 (Dec 2024) |
|---|---|---|
| GPQA Diamond | 91.7% | 53.8% |
| LMArena Expert | 1464 | 1254 |
| SimpleQA Verified | 52.9% | — |
| Confabulations | — | 20.1% |
| Vectara Hallucination Rate | 8.6% | — |
Multilingual DeepSeek V4 Pro leads
DeepSeek V4 Pro: 54.4 (#45), Grok-2 (Dec 2024): 43.1 (#188)
| Benchmark | DeepSeek V4 Pro | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Non-English | 1439 | 1282 |
| LMArena Chinese | 1486 | 1289 |
| LMArena French | 1472 | 1318 |
| LMArena German | 1458 | 1287 |
| LMArena Japanese | 1445 | 1244 |
| LMArena Korean | 1447 | 1237 |
| LMArena Russian | 1453 | 1286 |
| LMArena Spanish | 1458 | 1281 |
Instruction Following DeepSeek V4 Pro leads
DeepSeek V4 Pro: 76.1 (#47), Grok-2 (Dec 2024): 66.9 (#202)
| Benchmark | DeepSeek V4 Pro | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Instruction Following | 1448 | 1270 |
| LiveBench Instruction Following | — | 69.6% |
Long Context DeepSeek V4 Pro leads
DeepSeek V4 Pro: 45.0 (#51), Grok-2 (Dec 2024): 38.8 (#190)
| Benchmark | DeepSeek V4 Pro | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Longer Query | 1458 | 1276 |
| CL-bench Life | 13.5% | — |
Writing & Preference DeepSeek V4 Pro leads
DeepSeek V4 Pro: 65.5 (#46), Grok-2 (Dec 2024): 48.6 (#198)
| Benchmark | DeepSeek V4 Pro | Grok-2 (Dec 2024) |
|---|---|---|
| LMArena Text | 1451 | 1305 |
| LMArena Creative Writing | 1446 | 1284 |
| LMArena Multi-Turn | 1467 | 1290 |
| Short-Story Creative Writing | — | 63.6% |
| EQ-Bench Creative Writing | 1553 | — |
| EQ-Bench 4 | 1166 | — |
| LiveBench Language | — | 45.6% |
Frequently asked questions
Is DeepSeek V4 Pro better than Grok-2 (Dec 2024)?
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 33.7 on the Noometry Index.
Is DeepSeek V4 Pro or Grok-2 (Dec 2024) better for coding?
DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 33.3 in the Noometry coding category.
How many benchmarks do DeepSeek V4 Pro and Grok-2 (Dec 2024) share?
22 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and Grok-2 (Dec 2024) has 34.