Model comparison
DeepSeek-V3.1 vs Grok 4.6
Grok 4.6 is the stronger model overall, scoring 56.9 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 7.1× less per token, which makes it the better buy when Grok 4.6's lead doesn't matter for your workload.
Last verified . 22 shared benchmarks.
Summary
- They share 22 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 0 categories and Grok 4.6 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Grok 4.6 leads 61.4 to 27.9.
- The biggest single-benchmark swing is SimpleBench: 40% for DeepSeek-V3.1 and 75.9% for Grok 4.6.
- DeepSeek-V3.1 is cheaper at $0.25 / $0.95 per million input/output tokens, against $2 / $6 for Grok 4.6.
- Grok 4.6 accepts more context: 500K tokens versus 164K.
- DeepSeek-V3.1 has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-V3.1 | Grok 4.6 | |
|---|---|---|
| Provider | DeepSeek | xAI |
| Noometry Index | 42.8 | 56.9 |
| Released | 2025-08-21 | 2026-08-12 |
| Weights | Open | Proprietary |
| Context window | 164K | 500K |
| Max output | 8K | 500K |
| Input $ / M tokens | $0.25 | $2 |
| Output $ / M tokens | $0.95 | $6 |
| Results tracked | 27 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4.6 leads
DeepSeek-V3.1: 40.3 (#144), Grok 4.6: 58.5 (#16)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| WeirdML | 38.4% | 67.3% |
| LMArena Coding | 1417 | 1465 |
| DeepSWE | — | 67.5% |
| FrontierCode | — | 48% |
| CursorBench | — | 41.4% |
| LMArena WebDev | — | 1617 |
| FrontierSWE | — | 25.3% |
| SciCode | — | 56.5% |
| ALE-Bench | — | 1,508 |
Agentic & Tool Use Not comparable
DeepSeek-V3.1: —, Grok 4.6: 39.4 (#27)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| APEX-Agents | — | 65.3% |
| GDP.pdf | — | 17.2% |
| Vending-Bench 2 | — | 9,047 |
Reasoning Grok 4.6 leads
DeepSeek-V3.1: 27.9 (#110), Grok 4.6: 61.4 (#20)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| SimpleBench | 40% | 75.9% |
| LMArena Hard Prompts | 1417 | 1447 |
| DTBench | 82.7% | 97.3% |
| LMCA | 24.3% | 48.5% |
| Epoch Capabilities Index | 139.92 | 156.44 |
| ARC-AGI-2 | — | 67.1% |
| Kagi LLM Benchmark | 53.2% | — |
| NYT Connections (extended) | — | 80% |
| ARC-AGI-1 | — | 87.5% |
| CritPt | — | 19.7% |
| Chess Puzzles | — | 40% |
| EBR-Bench | — | 30.5% |
| Mystery Game Puzzles | — | 34% |
| ForecastBench | 58 | — |
Math Grok 4.6 leads
DeepSeek-V3.1: 38.9 (#122), Grok 4.6: 67.0 (#24)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| LMArena Math | 1420 | 1423 |
| FrontierMath (Tiers 1-3) | — | 66% |
| FrontierMath Tier 4 | — | 31.7% |
| OTIS Mock AIME 2024-2025 | — | 99.2% |
| ProofBench | — | 51% |
Knowledge Grok 4.6 leads
DeepSeek-V3.1: 43.7 (#90), Grok 4.6: 63.3 (#20)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| LMArena Expert | 1405 | 1467 |
| GPQA Diamond | — | 94% |
| SimpleQA Verified | — | 49.3% |
| Vectara Hallucination Rate | 5.5% | — |
Multimodal Not comparable
DeepSeek-V3.1: —, Grok 4.6: 43.6 (#23)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| LMArena Vision | — | 1263 |
| Blueprint-Bench 2 | — | 33.2% |
| Furniture Assembly | — | 40% |
| LMArena Document | — | 1452 |
Multilingual Grok 4.6 leads
DeepSeek-V3.1: 51.6 (#106), Grok 4.6: 53.0 (#74)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| LMArena Non-English | 1400 | 1420 |
| LMArena Chinese | 1469 | 1480 |
| LMArena French | 1447 | 1461 |
| LMArena German | 1411 | 1431 |
| LMArena Japanese | 1378 | 1376 |
| LMArena Korean | 1337 | 1397 |
| LMArena Russian | 1405 | 1422 |
| LMArena Spanish | 1431 | 1404 |
Instruction Following Grok 4.6 leads
DeepSeek-V3.1: 73.9 (#110), Grok 4.6: 75.4 (#63)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| LMArena Instruction Following | 1400 | 1431 |
Long Context Grok 4.6 leads
DeepSeek-V3.1: 36.3 (#232), Grok 4.6: 44.5 (#66)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| LMArena Longer Query | 1422 | 1454 |
| Fiction.LiveBench | 52.8% | — |
Writing & Preference Grok 4.6 leads
DeepSeek-V3.1: 60.3 (#98), Grok 4.6: 62.3 (#80)
| Benchmark | DeepSeek-V3.1 | Grok 4.6 |
|---|---|---|
| LMArena Text | 1420 | 1428 |
| LMArena Creative Writing | 1401 | 1428 |
| LMArena Multi-Turn | 1408 | 1425 |
| EQ-Bench Creative Writing | 1436 | — |
Frequently asked questions
Is DeepSeek-V3.1 better than Grok 4.6?
Grok 4.6 is the stronger model overall, scoring 56.9 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 7.1× less per token, which makes it the better buy when Grok 4.6's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3.1 or Grok 4.6?
DeepSeek-V3.1 is cheaper. It lists at $0.25 per million input tokens and $0.95 per million output tokens; Grok 4.6 lists at $2 and $6.
Is DeepSeek-V3.1 or Grok 4.6 better for coding?
Grok 4.6 scores higher on coding benchmarks: 58.5 versus 40.3 in the Noometry coding category.
Which has the bigger context window?
Grok 4.6 does, with 500K tokens against 164K.
How many benchmarks do DeepSeek-V3.1 and Grok 4.6 share?
22 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and Grok 4.6 has 49.