Model comparison
Grok 4 vs Llama 2-70B
Grok 4 is the stronger model overall, scoring 47.8 to 24.1 on the Noometry Index.
Last verified . 21 shared benchmarks.
MetaSummary
- They share 21 benchmarks with published results for both. Grok 4 scores higher in 8 categories and Llama 2-70B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Grok 4 leads 53.7 to 7.4.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 84% for Grok 4 and 0% for Llama 2-70B.
- Llama 2-70B has downloadable open weights; the other is API-only.
Side by side
| Grok 4 | Llama 2-70B | |
|---|---|---|
| Provider | xAI | Meta |
| Noometry Index | 47.8 | 24.1 |
| Released | 2025-07-09 | 2023-07-18 |
| Weights | Proprietary | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 48 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4 leads
Grok 4: 49.9 (#44), Llama 2-70B: 30.8 (#281)
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| LMArena Coding | 1408 | 1079 |
| Aider Polyglot | 79.6% | — |
| WeirdML | 45.7% | — |
Agentic & Tool Use Not comparable
Grok 4: 32.1 (#62), Llama 2-70B: —
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| Terminal-Bench | 27.2% | — |
| Berkeley Function Calling Leaderboard | 63% | — |
| GDPval | 21.1% | — |
| Cybench | 43% | — |
| DeepResearch Bench | 47.3% | — |
| BALROG | 43.6% | — |
| LMArena Search | 1142 | — |
| METR Time Horizons | 66.6% | — |
Reasoning Grok 4 leads
Grok 4: 36.9 (#59), Llama 2-70B: 13.7 (#324)
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| LMArena Hard Prompts | 1409 | 1073 |
| Epoch Capabilities Index | 146.46 | 113.59 |
| ForecastBench | 60.9 | 51.4 |
| ARC-AGI-2 | 16% | — |
| SimpleBench | 60.5% | — |
| Kagi LLM Benchmark | 73.6% | — |
| ARC-AGI-1 | 66.7% | — |
| Chess Puzzles | 28% | — |
| DTBench | — | 41.6% |
| BIG-Bench Hard | — | 64.9% |
| CommonsenseQA 2.0 | — | 50% |
| HellaSwag | — | 85.3% |
| LAMBADA | — | 78.9% |
| PIQA | — | 82.8% |
| WinoGrande | — | 80.2% |
Math Grok 4 leads
Grok 4: 46.5 (#58), Llama 2-70B: 7.5 (#320)
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 84% | 0% |
| LMArena Math | 1422 | 1091 |
| Omni-MATH | 60.3% | — |
| MATH Level 5 | — | 3.3% |
| FrontierMath (Feb 2025 set) | 19.7% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
| GSM8K | — | 69.6% |
Knowledge Grok 4 leads
Grok 4: 53.7 (#50), Llama 2-70B: 7.4 (#305)
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| GPQA Diamond | 87% | 26.3% |
| LMArena Expert | 1417 | 1039 |
| MMLU-Pro | 85.1% | — |
| Confabulations | 12.4% | — |
| GPQA (HELM) | 72.7% | — |
| ARC (AI2) Challenge | — | 78.3% |
| BoolQ | — | 88.6% |
| MMLU | — | 69.9% |
| OpenBookQA | — | 60.2% |
| TriviaQA | — | 87.6% |
Multimodal Not comparable
Grok 4: 33.1 (#89), Llama 2-70B: —
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| LMArena Vision | 1211 | — |
| GeoBench | 45% | — |
Multilingual Grok 4 leads
Grok 4: 52.0 (#98), Llama 2-70B: 27.9 (#269)
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| LMArena Non-English | 1403 | 1045 |
| LMArena Chinese | 1428 | 994 |
| LMArena French | 1418 | 1089 |
| LMArena German | 1429 | 1041 |
| LMArena Japanese | 1396 | 927 |
| LMArena Korean | 1376 | 964 |
| LMArena Russian | 1409 | 1083 |
| LMArena Spanish | 1419 | 1142 |
Instruction Following Grok 4 leads
Grok 4: 79.4 (#4), Llama 2-70B: 55.1 (#273)
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| LMArena Instruction Following | 1387 | 1071 |
| IFEval | 94.9% | — |
Long Context Grok 4 leads
Grok 4: 63.1 (#4), Llama 2-70B: 32.2 (#265)
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| LMArena Longer Query | 1409 | 1062 |
| Fiction.LiveBench | 94.4% | — |
Writing & Preference Grok 4 leads
Grok 4: 58.7 (#111), Llama 2-70B: 32.5 (#274)
| Benchmark | Grok 4 | Llama 2-70B |
|---|---|---|
| LMArena Text | 1412 | 1115 |
| LMArena Creative Writing | 1397 | 1075 |
| LMArena Multi-Turn | 1416 | 1088 |
| Short-Story Creative Writing | 76.9% | — |
| WildBench | 79.7% | — |
Frequently asked questions
Is Grok 4 better than Llama 2-70B?
Grok 4 is the stronger model overall, scoring 47.8 to 24.1 on the Noometry Index.
Is Grok 4 or Llama 2-70B better for coding?
Grok 4 scores higher on coding benchmarks: 49.9 versus 30.8 in the Noometry coding category.
How many benchmarks do Grok 4 and Llama 2-70B share?
21 benchmarks have published results for both models. Grok 4 has 48 scored results on Noometry and Llama 2-70B has 35.