# GLM-4.5V vs GPT-4 Turbo

> GLM-4.5V is the stronger model overall, scoring 39.8 to 30.5 on the Noometry Index.

- Canonical page: https://noometry.com/compare/glm-4-5v-vs-gpt-4-turbo
- Last updated: 2026-10-12
- Shared benchmarks: 14

## Summary

- They share 14 benchmarks with published results for both. GLM-4.5V scores higher in 9 categories and GPT-4 Turbo in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GLM-4.5V leads 37.4 to 9.0.
- GLM-4.5V is cheaper at $0.60 / $1.80 per million input/output tokens, against $10 / $30 for GPT-4 Turbo.
- GPT-4 Turbo accepts more context: 128K tokens versus 64K.
- GLM-4.5V has downloadable open weights; the other is API-only.

## Snapshot

| | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| Provider | Z.ai (Zhipu) | OpenAI |
| Noometry Index | 39.8 | 30.5 |
| Rank | 158 | 292 |
| Context | 64K | 128K |
| Input $/M | $0.60 | $10 |
| Output $/M | $1.80 | $30 |
| Weights | Open | Proprietary |

## Coding

- GLM-4.5V: 39.5 (#155)
- GPT-4 Turbo: 33.8 (#249)

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| LMArena Coding | 1347 | 1268 |
| WeirdML | — | 18% |
| BigCodeBench Instruct | — | 48.2% |
| BigCodeBench Complete | — | 58.2% |
| HumanEval+ | — | 86.6% |
| MBPP+ | — | 73.3% |

## Agentic & Tool Use

- GLM-4.5V: —
- GPT-4 Turbo: —

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| METR Time Horizons | — | 36.7% |

## Reasoning

- GLM-4.5V: 27.4 (#119)
- GPT-4 Turbo: 15.3 (#317)

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| LMArena Hard Prompts | 1334 | 1251 |
| SimpleBench | — | 25.1% |
| Kagi LLM Benchmark | 59.8% | — |
| Chess Puzzles | — | 6% |
| DTBench | — | 61.6% |
| LMCA | — | 9.8% |
| Epoch Capabilities Index | — | 127.25 |
| ForecastBench | — | 59.4 |

## Math

- GLM-4.5V: 37.4 (#159)
- GPT-4 Turbo: 9.0 (#322)

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| LMArena Math | 1354 | 1272 |
| FrontierMath (Tiers 1-3) | — | 0.7% |
| OTIS Mock AIME 2024-2025 | — | 6.7% |
| MATH Level 5 | — | 46.7% |

## Knowledge

- GLM-4.5V: 37.5 (#156)
- GPT-4 Turbo: 24.3 (#268)

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| LMArena Expert | 1353 | 1223 |
| GPQA Diamond | — | 46.6% |
| Confabulations | — | 28.4% |
| MMLU | — | 81.3% |

## Multimodal

- GLM-4.5V: 34.3 (#92)
- GPT-4 Turbo: 30.6 (#110)

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| LMArena Vision | 1154 | 1090 |

## Multilingual

- GLM-4.5V: 44.6 (#177)
- GPT-4 Turbo: 40.5 (#216)

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| LMArena Non-English | 1303 | 1245 |
| LMArena Chinese | 1337 | 1242 |
| LMArena Russian | 1298 | 1259 |
| LMArena Spanish | 1336 | 1260 |
| LMArena French | — | 1276 |
| LMArena German | — | 1259 |
| LMArena Japanese | — | 1194 |
| LMArena Korean | — | 1187 |

## Instruction Following

- GLM-4.5V: 69.2 (#175)
- GPT-4 Turbo: 65.8 (#216)

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| LMArena Instruction Following | 1311 | 1249 |

## Long Context

- GLM-4.5V: 39.6 (#171)
- GPT-4 Turbo: 38.0 (#206)

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| LMArena Longer Query | 1304 | 1254 |

## Writing & Preference

- GLM-4.5V: 52.5 (#170)
- GPT-4 Turbo: 47.7 (#206)

| Benchmark | GLM-4.5V | GPT-4 Turbo |
|---|---|---|
| LMArena Text | 1333 | 1272 |
| LMArena Creative Writing | 1295 | 1269 |
| LMArena Multi-Turn | 1332 | 1267 |

## FAQ

### Is GLM-4.5V better than GPT-4 Turbo?

GLM-4.5V is the stronger model overall, scoring 39.8 to 30.5 on the Noometry Index.

### Which is cheaper, GLM-4.5V or GPT-4 Turbo?

GLM-4.5V is cheaper. It lists at $0.60 per million input tokens and $1.80 per million output tokens; GPT-4 Turbo lists at $10 and $30.

### Is GLM-4.5V or GPT-4 Turbo better for coding?

GLM-4.5V scores higher on coding benchmarks: 39.5 versus 33.8 in the Noometry coding category.

### Which has the bigger context window?

GPT-4 Turbo does, with 128K tokens against 64K.

### How many benchmarks do GLM-4.5V and GPT-4 Turbo share?

14 benchmarks have published results for both models. GLM-4.5V has 15 scored results on Noometry and GPT-4 Turbo has 36.