Model comparison
DeepSeek-V2.5 (Sep 2024) vs Grok 4.20 (Non-Reasoning)
Grok 4.20 (Non-Reasoning) is the stronger model overall, scoring 48.6 to 37.6 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 0 categories and Grok 4.20 (Non-Reasoning) in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Grok 4.20 (Non-Reasoning) leads 52.3 to 25.6.
- DeepSeek-V2.5 (Sep 2024) has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) | |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 37.6 | 48.6 |
| Released | 2024-09-06 | 2026-02-17 |
| Weights | Open | Proprietary |
| Context window | — | 1M |
| Max output | — | 30K |
| Input $ / M tokens | — | $1.25 |
| Output $ / M tokens | — | $2.50 |
| Results tracked | 22 | 46 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4.20 (Non-Reasoning) leads
DeepSeek-V2.5 (Sep 2024): 31.7 (#281), Grok 4.20 (Non-Reasoning): 42.1 (#112)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Coding | 1309 | 1459 |
| Aider Polyglot | 17.8% | — |
| LMArena WebDev | — | 1375 |
| WeirdML | — | 52.3% |
| BigCodeBench Instruct | 48.6% | — |
| BigCodeBench Complete | 53.2% | — |
| ALE-Bench | — | 1,150 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |
Agentic & Tool Use Not comparable
DeepSeek-V2.5 (Sep 2024): —, Grok 4.20 (Non-Reasoning): 34.4 (#46)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| Terminal-Bench | — | 57.3% |
| τ²-bench Banking | — | 18% |
| LMArena Search | — | 1189 |
| Vending-Bench 2 | — | 4,663 |
Reasoning Grok 4.20 (Non-Reasoning) leads
DeepSeek-V2.5 (Sep 2024): 25.6 (#145), Grok 4.20 (Non-Reasoning): 52.3 (#32)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1451 |
| ARC-AGI-2 | — | 65.1% |
| Kagi LLM Benchmark | — | 75% |
| NYT Connections (extended) | — | 85.4% |
| ARC-AGI-1 | — | 89.5% |
| Chess Puzzles | — | 24% |
| Thematic Generalization | — | 63.8% |
| DTBench | — | 90.1% |
| LMCA | — | 38.7% |
| Epoch Capabilities Index | — | 151.98 |
| ForecastBench | — | 61.4 |
Math Grok 4.20 (Non-Reasoning) leads
DeepSeek-V2.5 (Sep 2024): 35.9 (#177), Grok 4.20 (Non-Reasoning): 48.2 (#65)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Math | 1288 | 1455 |
| FrontierMath (Tiers 1-3) | — | 44.9% |
| FrontierMath Tier 4 | — | 17.1% |
| OTIS Mock AIME 2024-2025 | — | 92.2% |
| ProofBench | — | 14% |
Knowledge Grok 4.20 (Non-Reasoning) leads
DeepSeek-V2.5 (Sep 2024): 34.8 (#193), Grok 4.20 (Non-Reasoning): 52.8 (#60)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Expert | 1266 | 1439 |
| GPQA Diamond | — | 89.3% |
| SimpleQA Verified | — | 30.2% |
Multimodal Not comparable
DeepSeek-V2.5 (Sep 2024): —, Grok 4.20 (Non-Reasoning): 33.3 (#98)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Vision | — | 1263 |
| Blueprint-Bench 2 | — | 0% |
| LMArena Document | — | 1416 |
Multilingual Grok 4.20 (Non-Reasoning) leads
DeepSeek-V2.5 (Sep 2024): 42.5 (#193), Grok 4.20 (Non-Reasoning): 54.5 (#40)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Non-English | 1273 | 1441 |
| LMArena Chinese | 1318 | 1481 |
| LMArena French | 1289 | 1476 |
| LMArena German | 1258 | 1465 |
| LMArena Japanese | 1228 | 1449 |
| LMArena Korean | 1209 | 1417 |
| LMArena Russian | 1289 | 1458 |
| LMArena Spanish | 1248 | 1443 |
Instruction Following Grok 4.20 (Non-Reasoning) leads
DeepSeek-V2.5 (Sep 2024): 67.5 (#194), Grok 4.20 (Non-Reasoning): 74.8 (#83)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Instruction Following | 1280 | 1420 |
Long Context Grok 4.20 (Non-Reasoning) leads
DeepSeek-V2.5 (Sep 2024): 39.5 (#174), Grok 4.20 (Non-Reasoning): 45.5 (#34)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Longer Query | 1301 | 1437 |
| CL-bench | — | 22.2% |
| CL-bench Life | — | 11.9% |
Writing & Preference Grok 4.20 (Non-Reasoning) leads
DeepSeek-V2.5 (Sep 2024): 49.8 (#187), Grok 4.20 (Non-Reasoning): 65.7 (#44)
| Benchmark | DeepSeek-V2.5 (Sep 2024) | Grok 4.20 (Non-Reasoning) |
|---|---|---|
| LMArena Text | 1294 | 1451 |
| LMArena Creative Writing | 1285 | 1438 |
| LMArena Multi-Turn | 1297 | 1456 |
| EQ-Bench Creative Writing | — | 1574 |
Frequently asked questions
Is DeepSeek-V2.5 (Sep 2024) better than Grok 4.20 (Non-Reasoning)?
Grok 4.20 (Non-Reasoning) is the stronger model overall, scoring 48.6 to 37.6 on the Noometry Index.
Is DeepSeek-V2.5 (Sep 2024) or Grok 4.20 (Non-Reasoning) better for coding?
Grok 4.20 (Non-Reasoning) scores higher on coding benchmarks: 42.1 versus 31.7 in the Noometry coding category.
How many benchmarks do DeepSeek-V2.5 (Sep 2024) and Grok 4.20 (Non-Reasoning) share?
17 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and Grok 4.20 (Non-Reasoning) has 46.