Model comparison
Grok 4.20 Multi-Agent vs Llama 3.1-70B
Grok 4.20 Multi-Agent is the stronger model overall, scoring 46.2 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 4.20 Multi-Agent's lead doesn't matter for your workload.
Last verified . 17 shared benchmarks.
MetaSummary
- They share 17 benchmarks with published results for both. Grok 4.20 Multi-Agent scores higher in 8 categories and Llama 3.1-70B in 0 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Grok 4.20 Multi-Agent leads 64.0 to 35.4.
- Llama 3.1-70B is cheaper at $0.40 / $0.40 per million input/output tokens, against $1.25 / $2.50 for Grok 4.20 Multi-Agent.
- Grok 4.20 Multi-Agent accepts more context: 1M tokens versus 128K.
- Llama 3.1-70B has downloadable open weights; the other is API-only.
Side by side
| Grok 4.20 Multi-Agent | Llama 3.1-70B | |
|---|---|---|
| Provider | xAI | Meta |
| Noometry Index | 46.2 | 29.6 |
| Released | 2026-03-09 | 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 | 20 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 43.0 (#92), Llama 3.1-70B: 30.3 (#296)
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| LMArena Coding | 1457 | 1260 |
| WeirdML | — | 9% |
| BigCodeBench Instruct | — | 46.1% |
| BigCodeBench Complete | — | 54.8% |
Agentic & Tool Use Not comparable
Grok 4.20 Multi-Agent: —, Llama 3.1-70B: 25.1 (#112)
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| TheAgentCompany | — | 6.9% |
| BALROG | — | 27.9% |
| LMArena Search | 1204 | — |
Reasoning Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 43.9 (#48), Llama 3.1-70B: 21.6 (#220)
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| LMArena Hard Prompts | 1448 | 1241 |
| NYT Connections (extended) | 89.6% | — |
| DTBench | — | 60% |
| LMCA | — | 14.8% |
| Epoch Capabilities Index | — | 125.92 |
Math Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 39.4 (#104), Llama 3.1-70B: 13.5 (#304)
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| LMArena Math | 1442 | 1252 |
| OTIS Mock AIME 2024-2025 | — | 3.6% |
| Omni-MATH | — | 21% |
| MATH Level 5 | — | 36.7% |
Knowledge Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 40.4 (#119), Llama 3.1-70B: 24.2 (#269)
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| LMArena Expert | 1445 | 1209 |
| GPQA Diamond | — | 44.2% |
| MMLU-Pro | — | 65.3% |
| GPQA (HELM) | — | 42.6% |
| MMLU | — | 80.1% |
Multimodal Not comparable
Grok 4.20 Multi-Agent: 40.5 (#48), Llama 3.1-70B: —
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| LMArena Vision | 1259 | — |
Multilingual Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 54.4 (#43), Llama 3.1-70B: 38.8 (#225)
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| LMArena Non-English | 1440 | 1219 |
| LMArena Chinese | 1475 | 1215 |
| LMArena French | 1466 | 1261 |
| LMArena German | 1456 | 1222 |
| LMArena Japanese | 1405 | 1132 |
| LMArena Korean | 1416 | 1140 |
| LMArena Russian | 1457 | 1234 |
| LMArena Spanish | 1447 | 1253 |
Instruction Following Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 74.8 (#84), Llama 3.1-70B: 65.3 (#223)
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| LMArena Instruction Following | 1420 | 1231 |
| IFEval | — | 82.1% |
Long Context Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 43.7 (#88), Llama 3.1-70B: 37.6 (#214)
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| LMArena Longer Query | 1431 | 1241 |
Writing & Preference Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 64.0 (#59), Llama 3.1-70B: 35.4 (#267)
| Benchmark | Grok 4.20 Multi-Agent | Llama 3.1-70B |
|---|---|---|
| LMArena Text | 1450 | 1261 |
| LMArena Creative Writing | 1436 | 1232 |
| LMArena Multi-Turn | 1452 | 1256 |
| EQ-Bench Creative Writing | — | 784 |
| WildBench | — | 75.8% |
Frequently asked questions
Is Grok 4.20 Multi-Agent better than Llama 3.1-70B?
Grok 4.20 Multi-Agent is the stronger model overall, scoring 46.2 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 4.20 Multi-Agent's lead doesn't matter for your workload.
Which is cheaper, Grok 4.20 Multi-Agent 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 4.20 Multi-Agent lists at $1.25 and $2.50.
Is Grok 4.20 Multi-Agent or Llama 3.1-70B better for coding?
Grok 4.20 Multi-Agent scores higher on coding benchmarks: 43.0 versus 30.3 in the Noometry coding category.
Which has the bigger context window?
Grok 4.20 Multi-Agent does, with 1M tokens against 128K.
How many benchmarks do Grok 4.20 Multi-Agent and Llama 3.1-70B share?
17 benchmarks have published results for both models. Grok 4.20 Multi-Agent has 20 scored results on Noometry and Llama 3.1-70B has 35.