# Gemma 4 31B IT vs GPT-5.4

> GPT-5.4 is the stronger model overall, scoring 59.5 to 43.2 on the Noometry Index. Gemma 4 31B IT costs 37× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gemma-4-31b-it-vs-gpt-5-4
- Last updated: 2026-10-10
- Shared benchmarks: 34

## Summary

- They share 34 benchmarks with published results for both. Gemma 4 31B IT scores higher in 0 categories and GPT-5.4 in 9 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.4 leads 61.7 to 26.4.
- The biggest single-benchmark swing is Chess Puzzles: 5% for Gemma 4 31B IT and 44% for GPT-5.4.
- Gemma 4 31B IT is cheaper at $0.09 / $0.34 per million input/output tokens, against $2.50 / $15 for GPT-5.4.
- GPT-5.4 accepts more context: 1.05M tokens versus 262K.
- Gemma 4 31B IT has downloadable open weights; the other is API-only.

## Snapshot

| | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| Provider | Google | OpenAI |
| Noometry Index | 43.2 | 59.5 |
| Rank | 85 | 14 |
| Context | 262K | 1.05M |
| Input $/M | $0.09 | $2.50 |
| Output $/M | $0.34 | $15 |
| Weights | Open | Proprietary |

## Coding

- Gemma 4 31B IT: 41.9 (#98)
- GPT-5.4: 52.8 (#28)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| LMArena WebDev | 1364 | 1465 |
| SciCode | 43.4% | 56.6% |
| WeirdML | 52.3% | 77.7% |
| LMArena Coding | 1459 | 1497 |
| ALE-Bench | 925.5 | 1,607 |
| SWE-bench Verified | — | 76.9% |
| DeepSWE | — | 51.8% |
| GSO | — | 31.4% |
| MirrorCode | — | 15.6% |
| AlgoTune | — | 1.85 |

## Agentic & Tool Use

- Gemma 4 31B IT: —
- GPT-5.4: 46.5 (#12)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| Terminal-Bench | — | 81.8% |
| APEX-Agents | — | 52.4% |
| τ²-bench Banking | — | 39.4% |
| DeepResearch Bench | — | 35.1% |
| PostTrainBench | — | 19% |
| GBAEval | — | 45.1% |
| LMArena Search | — | 1197 |
| METR Time Horizons | — | 74.3% |
| Vending-Bench 2 | — | 6,144 |

## Reasoning

- Gemma 4 31B IT: 26.4 (#119)
- GPT-5.4: 61.7 (#17)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| Kagi LLM Benchmark | 63.5% | 63.8% |
| NYT Connections (extended) | 70.6% | 91.3% |
| CritPt | 1.4% | 23.4% |
| Chess Puzzles | 5% | 44% |
| Thematic Generalization | 53% | 80% |
| LMArena Hard Prompts | 1448 | 1485 |
| DTBench | 82.7% | 94.4% |
| LMCA | 39.3% | 52% |
| Epoch Capabilities Index | 142.74 | 156.92 |
| ARC-AGI-2 | — | 74% |
| ARC-AGI-1 | — | 93.7% |
| EnigmaEval | — | 16% |
| EBR-Bench | — | 25.4% |
| Mystery Game Puzzles | — | 37% |
| Surface Evolver Bench | 30.6% | — |
| ForecastBench | — | 59.5 |

## Math

- Gemma 4 31B IT: 41.3 (#72)
- GPT-5.4: 73.3 (#14)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 73.3% | 97.8% |
| LMArena Math | 1464 | 1488 |
| FrontierMath (Tiers 1-3) | — | 78.6% |
| FrontierMath Tier 4 | — | 49% |
| MathArena Final-Answer Competitions | — | 83.1% |
| ProofBench | — | 56% |
| FrontierMath (Feb 2025 set) | — | 47.6% |
| FrontierMath Tier 4 (v1) | — | 27.1% |

## Knowledge

- Gemma 4 31B IT: 37.9 (#141)
- GPT-5.4: 65.3 (#13)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| GPQA Diamond | 75.8% | 93.3% |
| SimpleQA Verified | 10.4% | 45.1% |
| Vectara Hallucination Rate | 7.4% | 7% |
| LMArena Expert | 1465 | 1507 |
| Humanity's Last Exam | — | 36.2% |

## Multimodal

- Gemma 4 31B IT: 41.0 (#32)
- GPT-5.4: 43.8 (#19)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| LMArena Vision | 1278 | 1302 |
| LMArena Document | 1425 | 1471 |
| Blueprint-Bench 2 | — | 27.1% |
| Furniture Assembly | — | 37.5% |

## Multilingual

- Gemma 4 31B IT: 54.0 (#52)
- GPT-5.4: 56.4 (#22)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| LMArena Non-English | 1431 | 1465 |
| LMArena Chinese | 1478 | 1520 |
| LMArena French | 1436 | 1493 |
| LMArena Russian | 1462 | 1480 |
| LMArena Spanish | 1443 | 1453 |
| LMArena German | — | 1472 |
| LMArena Japanese | — | 1484 |
| LMArena Korean | — | 1449 |

## Instruction Following

- Gemma 4 31B IT: 75.6 (#58)
- GPT-5.4: 77.3 (#25)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| LMArena Instruction Following | 1433 | 1469 |

## Long Context

- Gemma 4 31B IT: 44.2 (#67)
- GPT-5.4: 50.3 (#8)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| LMArena Longer Query | 1446 | 1474 |
| CL-bench | — | 27.9% |
| CL-bench Life | — | 21.7% |

## Writing & Preference

- Gemma 4 31B IT: 60.7 (#92)
- GPT-5.4: 72.1 (#18)

| Benchmark | Gemma 4 31B IT | GPT-5.4 |
|---|---|---|
| LMArena Text | 1443 | 1469 |
| LMArena Creative Writing | 1416 | 1439 |
| EQ-Bench Creative Writing | 1368 | 1840 |
| EQ-Bench 4 | 1120 | 1272 |
| LMArena Multi-Turn | 1452 | 1482 |

## FAQ

### Is Gemma 4 31B IT better than GPT-5.4?

GPT-5.4 is the stronger model overall, scoring 59.5 to 43.2 on the Noometry Index. Gemma 4 31B IT costs 37× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.

### Which is cheaper, Gemma 4 31B IT or GPT-5.4?

Gemma 4 31B IT is cheaper. It lists at $0.09 per million input tokens and $0.34 per million output tokens; GPT-5.4 lists at $2.50 and $15.

### Is Gemma 4 31B IT or GPT-5.4 better for coding?

GPT-5.4 scores higher on coding benchmarks: 52.8 versus 41.9 in the Noometry coding category.

### Which has the bigger context window?

GPT-5.4 does, with 1.05M tokens against 262K.

### How many benchmarks do Gemma 4 31B IT and GPT-5.4 share?

34 benchmarks have published results for both models. Gemma 4 31B IT has 35 scored results on Noometry and GPT-5.4 has 68.