Model comparison
Grok 4.3 vs Llama 3.1-70B
Grok 4.3 is the stronger model overall, scoring 43.8 to 29.6 on the Noometry Index. Llama 3.1-70B costs 3.9× less per token, which makes it the better buy when Grok 4.3's lead doesn't matter for your workload.
Last verified . 23 shared benchmarks.
MetaSummary
- They share 23 benchmarks with published results for both. Grok 4.3 scores higher in 9 categories and Llama 3.1-70B in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Grok 4.3 leads 46.0 to 13.5.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 93.3% for Grok 4.3 and 3.6% for Llama 3.1-70B.
- Llama 3.1-70B is cheaper at $0.40 / $0.40 per million input/output tokens, against $1.25 / $2.50 for Grok 4.3.
- Grok 4.3 accepts more context: 1M tokens versus 128K.
- Llama 3.1-70B has downloadable open weights; the other is API-only.
Side by side
| Grok 4.3 | Llama 3.1-70B | |
|---|---|---|
| Provider | xAI | Meta |
| Noometry Index | 43.8 | 29.6 |
| Released | 2026-04-17 | 2024-07-23 |
| Weights | Proprietary | Open |
| Context window | 1M | 128K |
| Max output | 30K | 4K |
| Input $ / M tokens | $1.25 | $0.40 |
| Output $ / M tokens | $2.50 | $0.40 |
| Results tracked | 40 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4.3 leads
Grok 4.3: 41.6 (#121), Llama 3.1-70B: 30.3 (#296)
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| WeirdML | 49.9% | 9% |
| LMArena Coding | 1415 | 1260 |
| LMArena WebDev | 1357 | — |
| SciCode | 47.3% | — |
| BigCodeBench Instruct | — | 46.1% |
| BigCodeBench Complete | — | 54.8% |
| ALE-Bench | 944.17 | — |
Agentic & Tool Use Grok 4.3 leads
Grok 4.3: 27.7 (#99), Llama 3.1-70B: 25.1 (#112)
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| TheAgentCompany | — | 6.9% |
| BALROG | — | 27.9% |
| GDP.pdf | 8% | — |
| LMArena Search | 1165 | — |
| Vending-Bench 2 | 35.26 | — |
Reasoning Grok 4.3 leads
Grok 4.3: 35.9 (#68), Llama 3.1-70B: 21.6 (#220)
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| LMArena Hard Prompts | 1396 | 1241 |
| DTBench | 90.7% | 60% |
| LMCA | 38.3% | 14.8% |
| Epoch Capabilities Index | 149.16 | 125.92 |
| NYT Connections (extended) | 55.2% | — |
| CritPt | 8% | — |
| Chess Puzzles | 25% | — |
| ForecastBench | 60.3 | — |
Math Grok 4.3 leads
Grok 4.3: 46.0 (#74), Llama 3.1-70B: 13.5 (#304)
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 93.3% | 3.6% |
| LMArena Math | 1388 | 1252 |
| FrontierMath (Tiers 1-3) | 42.8% | — |
| FrontierMath Tier 4 | 14.6% | — |
| ProofBench | 11% | — |
| Omni-MATH | — | 21% |
| MATH Level 5 | — | 36.7% |
Knowledge Grok 4.3 leads
Grok 4.3: 52.5 (#62), Llama 3.1-70B: 24.2 (#269)
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| GPQA Diamond | 88.8% | 44.2% |
| LMArena Expert | 1385 | 1209 |
| SimpleQA Verified | 33.2% | — |
| MMLU-Pro | — | 65.3% |
| GPQA (HELM) | — | 42.6% |
| MMLU | — | 80.1% |
Multimodal Not comparable
Grok 4.3: 31.6 (#104), Llama 3.1-70B: —
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| LMArena Vision | 1229 | — |
| Blueprint-Bench 2 | 0% | — |
Multilingual Grok 4.3 leads
Grok 4.3: 50.5 (#120), Llama 3.1-70B: 38.8 (#225)
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| LMArena Non-English | 1385 | 1219 |
| LMArena Chinese | 1422 | 1215 |
| LMArena French | 1412 | 1261 |
| LMArena German | 1395 | 1222 |
| LMArena Japanese | 1379 | 1132 |
| LMArena Korean | 1356 | 1140 |
| LMArena Russian | 1399 | 1234 |
| LMArena Spanish | 1398 | 1253 |
Instruction Following Grok 4.3 leads
Grok 4.3: 72.1 (#140), Llama 3.1-70B: 65.3 (#223)
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| LMArena Instruction Following | 1366 | 1231 |
| IFEval | — | 82.1% |
Long Context Grok 4.3 leads
Grok 4.3: 42.5 (#123), Llama 3.1-70B: 37.6 (#214)
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| LMArena Longer Query | 1393 | 1241 |
Writing & Preference Grok 4.3 leads
Grok 4.3: 58.5 (#118), Llama 3.1-70B: 35.4 (#267)
| Benchmark | Grok 4.3 | Llama 3.1-70B |
|---|---|---|
| LMArena Text | 1397 | 1261 |
| LMArena Creative Writing | 1380 | 1232 |
| LMArena Multi-Turn | 1406 | 1256 |
| EQ-Bench Creative Writing | — | 784 |
| WildBench | — | 75.8% |
| EQ-Bench 4 | 1075 | — |
Frequently asked questions
Is Grok 4.3 better than Llama 3.1-70B?
Grok 4.3 is the stronger model overall, scoring 43.8 to 29.6 on the Noometry Index. Llama 3.1-70B costs 3.9× less per token, which makes it the better buy when Grok 4.3's lead doesn't matter for your workload.
Which is cheaper, Grok 4.3 or Llama 3.1-70B?
Llama 3.1-70B is cheaper. It lists at $0.40 per million input tokens and $0.40 per million output tokens; Grok 4.3 lists at $1.25 and $2.50.
Is Grok 4.3 or Llama 3.1-70B better for coding?
Grok 4.3 scores higher on coding benchmarks: 41.6 versus 30.3 in the Noometry coding category.
Which has the bigger context window?
Grok 4.3 does, with 1M tokens against 128K.
How many benchmarks do Grok 4.3 and Llama 3.1-70B share?
23 benchmarks have published results for both models. Grok 4.3 has 40 scored results on Noometry and Llama 3.1-70B has 35.