# Grok 4.20 (Non-Reasoning) vs Grok 4.5

> Grok 4.5 is the stronger model overall, scoring 55.0 to 48.5 on the Noometry Index. Grok 4.20 (Non-Reasoning) costs 1.9× less per token, which makes it the better buy when Grok 4.5's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/grok-4-20-vs-grok-4-5
- Last updated: 2026-10-10
- Shared benchmarks: 41

## Summary

- They share 41 benchmarks with published results for both. Grok 4.20 (Non-Reasoning) scores higher in 2 categories and Grok 4.5 in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where Grok 4.5 leads 60.7 to 48.0.
- The biggest single-benchmark swing is τ²-bench Banking: 18% for Grok 4.20 (Non-Reasoning) and 47.9% for Grok 4.5.
- Grok 4.20 (Non-Reasoning) is cheaper at $1.25 / $2.50 per million input/output tokens, against $2 / $6 for Grok 4.5.
- Grok 4.20 (Non-Reasoning) accepts more context: 1M tokens versus 500K.

## Snapshot

| | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| Provider | xAI | xAI |
| Noometry Index | 48.5 | 55.0 |
| Rank | 48 | 22 |
| Context | 1M | 500K |
| Input $/M | $1.25 | $2 |
| Output $/M | $2.50 | $6 |
| Weights | Proprietary | Proprietary |

## Coding

- Grok 4.20 (Non-Reasoning): 41.8 (#104)
- Grok 4.5: 51.7 (#34)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| LMArena WebDev | 1374 | 1552 |
| WeirdML | 52.3% | 46.4% |
| LMArena Coding | 1458 | 1473 |
| ALE-Bench | 1,150 | 1,309 |
| DeepSWE | — | 53.8% |
| FrontierCode | — | 42.4% |
| SciCode | — | 54.1% |

## Agentic & Tool Use

- Grok 4.20 (Non-Reasoning): 34.3 (#43)
- Grok 4.5: 44.4 (#15)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| τ²-bench Banking | 18% | 47.9% |
| LMArena Search | 1189 | 1213 |
| Vending-Bench 2 | 4,663 | 3,887 |
| Terminal-Bench | 57.3% | — |
| APEX-Agents | — | 56.2% |
| PostTrainBench | — | 23.4% |
| GBAEval | — | 65.4% |
| GDP.pdf | — | 14% |

## Reasoning

- Grok 4.20 (Non-Reasoning): 52.2 (#30)
- Grok 4.5: 56.0 (#23)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| ARC-AGI-2 | 65.1% | 52.6% |
| Kagi LLM Benchmark | 75% | 83.5% |
| NYT Connections (extended) | 85.4% | 79.9% |
| ARC-AGI-1 | 89.5% | 87.2% |
| Chess Puzzles | 24% | 36% |
| LMArena Hard Prompts | 1451 | 1462 |
| DTBench | 90.1% | 96.5% |
| LMCA | 38.7% | 45.2% |
| Epoch Capabilities Index | 152.01 | 154.02 |
| SimpleBench | — | 70% |
| CritPt | — | 15.4% |
| Thematic Generalization | 63.8% | — |
| Surface Evolver Bench | — | 74.4% |
| ForecastBench | 61.4 | — |

## Math

- Grok 4.20 (Non-Reasoning): 48.0 (#54)
- Grok 4.5: 60.7 (#29)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 44.9% | 57.2% |
| FrontierMath Tier 4 | 17.1% | 24.4% |
| OTIS Mock AIME 2024-2025 | 92.2% | 97.8% |
| ProofBench | 14% | 31% |
| LMArena Math | 1455 | 1457 |

## Knowledge

- Grok 4.20 (Non-Reasoning): 52.8 (#54)
- Grok 4.5: 62.3 (#21)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| GPQA Diamond | 89.3% | 93.4% |
| SimpleQA Verified | 30.2% | 48.3% |
| LMArena Expert | 1440 | 1466 |

## Multimodal

- Grok 4.20 (Non-Reasoning): 33.2 (#88)
- Grok 4.5: 37.8 (#63)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| LMArena Vision | 1263 | 1287 |
| Blueprint-Bench 2 | 0% | 27.3% |
| LMArena Document | 1416 | 1452 |
| Furniture Assembly | — | 22.5% |

## Multilingual

- Grok 4.20 (Non-Reasoning): 54.7 (#38)
- Grok 4.5: 54.5 (#43)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| LMArena Non-English | 1440 | 1439 |
| LMArena Chinese | 1481 | 1497 |
| LMArena French | 1477 | 1458 |
| LMArena German | 1463 | 1448 |
| LMArena Japanese | 1450 | 1425 |
| LMArena Korean | 1417 | 1402 |
| LMArena Russian | 1457 | 1447 |
| LMArena Spanish | 1442 | 1452 |

## Instruction Following

- Grok 4.20 (Non-Reasoning): 75.0 (#79)
- Grok 4.5: 76.3 (#45)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| LMArena Instruction Following | 1420 | 1446 |

## Long Context

- Grok 4.20 (Non-Reasoning): 45.4 (#33)
- Grok 4.5: 44.8 (#54)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| LMArena Longer Query | 1436 | 1463 |
| CL-bench | 22.2% | — |
| CL-bench Life | 11.9% | — |

## Writing & Preference

- Grok 4.20 (Non-Reasoning): 65.9 (#42)
- Grok 4.5: 65.9 (#40)

| Benchmark | Grok 4.20 (Non-Reasoning) | Grok 4.5 |
|---|---|---|
| LMArena Text | 1451 | 1448 |
| LMArena Creative Writing | 1439 | 1442 |
| EQ-Bench Creative Writing | 1574 | 1579 |
| LMArena Multi-Turn | 1456 | 1457 |

## FAQ

### Is Grok 4.20 (Non-Reasoning) better than Grok 4.5?

Grok 4.5 is the stronger model overall, scoring 55.0 to 48.5 on the Noometry Index. Grok 4.20 (Non-Reasoning) costs 1.9× less per token, which makes it the better buy when Grok 4.5's lead doesn't matter for your workload.

### Which is cheaper, Grok 4.20 (Non-Reasoning) or Grok 4.5?

Grok 4.20 (Non-Reasoning) is cheaper. It lists at $1.25 per million input tokens and $2.50 per million output tokens; Grok 4.5 lists at $2 and $6.

### Is Grok 4.20 (Non-Reasoning) or Grok 4.5 better for coding?

Grok 4.5 scores higher on coding benchmarks: 51.7 versus 41.8 in the Noometry coding category.

### Which has the bigger context window?

Grok 4.20 (Non-Reasoning) does, with 1M tokens against 500K.

### How many benchmarks do Grok 4.20 (Non-Reasoning) and Grok 4.5 share?

41 benchmarks have published results for both models. Grok 4.20 (Non-Reasoning) has 46 scored results on Noometry and Grok 4.5 has 52.