Model comparison
Gemma 4 31B IT vs Grok 4.5
Grok 4.5 is the stronger model overall, scoring 55.0 to 43.2 on the Noometry Index. Gemma 4 31B IT costs 20× less per token, which makes it the better buy when Grok 4.5's lead doesn't matter for your workload.
Last verified . 32 shared benchmarks.
GoogleSummary
- They share 32 benchmarks with published results for both. Gemma 4 31B IT scores higher in 1 category and Grok 4.5 in 8 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Grok 4.5 leads 56.0 to 26.4.
- The biggest single-benchmark swing is Surface Evolver Bench: 30.6% for Gemma 4 31B IT and 74.4% for Grok 4.5.
- Gemma 4 31B IT is cheaper at $0.09 / $0.34 per million input/output tokens, against $2 / $6 for Grok 4.5.
- Grok 4.5 accepts more context: 500K tokens versus 262K.
- Gemma 4 31B IT has downloadable open weights; the other is API-only.
Side by side
| Gemma 4 31B IT | Grok 4.5 | |
|---|---|---|
| Provider | xAI | |
| Noometry Index | 43.2 | 55.0 |
| Released | 2026-04-02 | 2026-07-08 |
| Weights | Open | Proprietary |
| Context window | 262K | 500K |
| Max output | 33K | 500K |
| Input $ / M tokens | $0.09 | $2 |
| Output $ / M tokens | $0.34 | $6 |
| Results tracked | 35 | 52 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 4.5 leads
Gemma 4 31B IT: 41.9 (#98), Grok 4.5: 51.7 (#34)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| LMArena WebDev | 1364 | 1552 |
| SciCode | 43.4% | 54.1% |
| WeirdML | 52.3% | 46.4% |
| LMArena Coding | 1459 | 1473 |
| ALE-Bench | 925.5 | 1,309 |
| DeepSWE | — | 53.8% |
| FrontierCode | — | 42.4% |
Agentic & Tool Use Not comparable
Gemma 4 31B IT: —, Grok 4.5: 44.4 (#15)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| APEX-Agents | — | 56.2% |
| τ²-bench Banking | — | 47.9% |
| PostTrainBench | — | 23.4% |
| GBAEval | — | 65.4% |
| GDP.pdf | — | 14% |
| LMArena Search | — | 1213 |
| Vending-Bench 2 | — | 3,887 |
Reasoning Grok 4.5 leads
Gemma 4 31B IT: 26.4 (#119), Grok 4.5: 56.0 (#23)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| Kagi LLM Benchmark | 63.5% | 83.5% |
| NYT Connections (extended) | 70.6% | 79.9% |
| CritPt | 1.4% | 15.4% |
| Chess Puzzles | 5% | 36% |
| LMArena Hard Prompts | 1448 | 1462 |
| DTBench | 82.7% | 96.5% |
| LMCA | 39.3% | 45.2% |
| Surface Evolver Bench | 30.6% | 74.4% |
| Epoch Capabilities Index | 142.74 | 154.02 |
| ARC-AGI-2 | — | 52.6% |
| SimpleBench | — | 70% |
| ARC-AGI-1 | — | 87.2% |
| Thematic Generalization | 53% | — |
Math Grok 4.5 leads
Gemma 4 31B IT: 41.3 (#72), Grok 4.5: 60.7 (#29)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 73.3% | 97.8% |
| LMArena Math | 1464 | 1457 |
| FrontierMath (Tiers 1-3) | — | 57.2% |
| FrontierMath Tier 4 | — | 24.4% |
| ProofBench | — | 31% |
Knowledge Grok 4.5 leads
Gemma 4 31B IT: 37.9 (#141), Grok 4.5: 62.3 (#21)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| GPQA Diamond | 75.8% | 93.4% |
| SimpleQA Verified | 10.4% | 48.3% |
| LMArena Expert | 1465 | 1466 |
| Vectara Hallucination Rate | 7.4% | — |
Multimodal Gemma 4 31B IT leads
Gemma 4 31B IT: 41.0 (#32), Grok 4.5: 37.8 (#63)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| LMArena Vision | 1278 | 1287 |
| LMArena Document | 1425 | 1452 |
| Blueprint-Bench 2 | — | 27.3% |
| Furniture Assembly | — | 22.5% |
Multilingual Too close to call
Gemma 4 31B IT: 54.0 (#52), Grok 4.5: 54.5 (#43)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| LMArena Non-English | 1431 | 1439 |
| LMArena Chinese | 1478 | 1497 |
| LMArena French | 1436 | 1458 |
| LMArena Russian | 1462 | 1447 |
| LMArena Spanish | 1443 | 1452 |
| LMArena German | — | 1448 |
| LMArena Japanese | — | 1425 |
| LMArena Korean | — | 1402 |
Instruction Following Too close to call
Gemma 4 31B IT: 75.6 (#58), Grok 4.5: 76.3 (#45)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| LMArena Instruction Following | 1433 | 1446 |
Long Context Too close to call
Gemma 4 31B IT: 44.2 (#67), Grok 4.5: 44.8 (#54)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| LMArena Longer Query | 1446 | 1463 |
Writing & Preference Grok 4.5 leads
Gemma 4 31B IT: 60.7 (#92), Grok 4.5: 65.9 (#40)
| Benchmark | Gemma 4 31B IT | Grok 4.5 |
|---|---|---|
| LMArena Text | 1443 | 1448 |
| LMArena Creative Writing | 1416 | 1442 |
| EQ-Bench Creative Writing | 1368 | 1579 |
| LMArena Multi-Turn | 1452 | 1457 |
| EQ-Bench 4 | 1120 | — |
Frequently asked questions
Is Gemma 4 31B IT better than Grok 4.5?
Grok 4.5 is the stronger model overall, scoring 55.0 to 43.2 on the Noometry Index. Gemma 4 31B IT costs 20× less per token, which makes it the better buy when Grok 4.5's lead doesn't matter for your workload.
Which is cheaper, Gemma 4 31B IT or Grok 4.5?
Gemma 4 31B IT is cheaper. It lists at $0.09 per million input tokens and $0.34 per million output tokens; Grok 4.5 lists at $2 and $6.
Is Gemma 4 31B IT or Grok 4.5 better for coding?
Grok 4.5 scores higher on coding benchmarks: 51.7 versus 41.9 in the Noometry coding category.
Which has the bigger context window?
Grok 4.5 does, with 500K tokens against 262K.
How many benchmarks do Gemma 4 31B IT and Grok 4.5 share?
32 benchmarks have published results for both models. Gemma 4 31B IT has 35 scored results on Noometry and Grok 4.5 has 52.