Model comparison
Grok 4.20 Multi-Agent vs Grok 4.6
Grok 4.6 is the stronger model overall, scoring 56.9 to 46.2 on the Noometry Index. Grok 4.20 Multi-Agent costs 1.9× less per token, which makes it the better buy when Grok 4.6's lead doesn't matter for your workload.
Last verified . 19 shared benchmarks.
Summary
- They share 19 benchmarks with published results for both. Grok 4.20 Multi-Agent scores higher in 2 categories and Grok 4.6 in 7 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Grok 4.6 leads 67.0 to 39.4.
- The biggest single-benchmark swing is NYT Connections (extended): 89.6% for Grok 4.20 Multi-Agent and 80% for Grok 4.6.
- Grok 4.20 Multi-Agent is cheaper at $1.25 / $2.50 per million input/output tokens, against $2 / $6 for Grok 4.6.
- Grok 4.20 Multi-Agent accepts more context: 1M tokens versus 500K.
Side by side
| Grok 4.20 Multi-Agent | Grok 4.6 | |
|---|---|---|
| Provider | xAI | xAI |
| Noometry Index | 46.2 | 56.9 |
| Released | 2026-03-09 | 2026-08-12 |
| Weights | Proprietary | Proprietary |
| Context window | 1M | 500K |
| Max output | 30K | 500K |
| Input $ / M tokens | $1.25 | $2 |
| Output $ / M tokens | $2.50 | $6 |
| Results tracked | 20 | 49 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4.6 leads
Grok 4.20 Multi-Agent: 43.0 (#92), Grok 4.6: 58.5 (#16)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| LMArena Coding | 1457 | 1465 |
| DeepSWE | — | 67.5% |
| FrontierCode | — | 48% |
| CursorBench | — | 41.4% |
| LMArena WebDev | — | 1617 |
| FrontierSWE | — | 25.3% |
| SciCode | — | 56.5% |
| WeirdML | — | 67.3% |
| ALE-Bench | — | 1,508 |
Agentic & Tool Use Not comparable
Grok 4.20 Multi-Agent: —, Grok 4.6: 39.4 (#27)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| APEX-Agents | — | 65.3% |
| GDP.pdf | — | 17.2% |
| LMArena Search | 1204 | — |
| Vending-Bench 2 | — | 9,047 |
Reasoning Grok 4.6 leads
Grok 4.20 Multi-Agent: 43.9 (#48), Grok 4.6: 61.4 (#20)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| NYT Connections (extended) | 89.6% | 80% |
| LMArena Hard Prompts | 1448 | 1447 |
| ARC-AGI-2 | — | 67.1% |
| SimpleBench | — | 75.9% |
| ARC-AGI-1 | — | 87.5% |
| CritPt | — | 19.7% |
| Chess Puzzles | — | 40% |
| EBR-Bench | — | 30.5% |
| Mystery Game Puzzles | — | 34% |
| DTBench | — | 97.3% |
| LMCA | — | 48.5% |
| Epoch Capabilities Index | — | 156.44 |
Math Grok 4.6 leads
Grok 4.20 Multi-Agent: 39.4 (#104), Grok 4.6: 67.0 (#24)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| LMArena Math | 1442 | 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
Grok 4.20 Multi-Agent: 40.4 (#119), Grok 4.6: 63.3 (#20)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| LMArena Expert | 1445 | 1467 |
| GPQA Diamond | — | 94% |
| SimpleQA Verified | — | 49.3% |
Multimodal Grok 4.6 leads
Grok 4.20 Multi-Agent: 40.5 (#48), Grok 4.6: 43.6 (#23)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| LMArena Vision | 1259 | 1263 |
| Blueprint-Bench 2 | — | 33.2% |
| Furniture Assembly | — | 40% |
| LMArena Document | — | 1452 |
Multilingual Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 54.4 (#43), Grok 4.6: 53.0 (#74)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| LMArena Non-English | 1440 | 1420 |
| LMArena Chinese | 1475 | 1480 |
| LMArena French | 1466 | 1461 |
| LMArena German | 1456 | 1431 |
| LMArena Japanese | 1405 | 1376 |
| LMArena Korean | 1416 | 1397 |
| LMArena Russian | 1457 | 1422 |
| LMArena Spanish | 1447 | 1404 |
Instruction Following Too close to call
Grok 4.20 Multi-Agent: 74.8 (#84), Grok 4.6: 75.4 (#63)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| LMArena Instruction Following | 1420 | 1431 |
Long Context Too close to call
Grok 4.20 Multi-Agent: 43.7 (#88), Grok 4.6: 44.5 (#66)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| LMArena Longer Query | 1431 | 1454 |
Writing & Preference Grok 4.20 Multi-Agent leads
Grok 4.20 Multi-Agent: 64.0 (#59), Grok 4.6: 62.3 (#80)
| Benchmark | Grok 4.20 Multi-Agent | Grok 4.6 |
|---|---|---|
| LMArena Text | 1450 | 1428 |
| LMArena Creative Writing | 1436 | 1428 |
| LMArena Multi-Turn | 1452 | 1425 |
Frequently asked questions
Is Grok 4.20 Multi-Agent better than Grok 4.6?
Grok 4.6 is the stronger model overall, scoring 56.9 to 46.2 on the Noometry Index. Grok 4.20 Multi-Agent costs 1.9× less per token, which makes it the better buy when Grok 4.6's lead doesn't matter for your workload.
Which is cheaper, Grok 4.20 Multi-Agent or Grok 4.6?
Grok 4.20 Multi-Agent is cheaper. It lists at $1.25 per million input tokens and $2.50 per million output tokens; Grok 4.6 lists at $2 and $6.
Is Grok 4.20 Multi-Agent or Grok 4.6 better for coding?
Grok 4.6 scores higher on coding benchmarks: 58.5 versus 43.0 in the Noometry coding category.
Which has the bigger context window?
Grok 4.20 Multi-Agent does, with 1M tokens against 500K.
How many benchmarks do Grok 4.20 Multi-Agent and Grok 4.6 share?
19 benchmarks have published results for both models. Grok 4.20 Multi-Agent has 20 scored results on Noometry and Grok 4.6 has 49.