# DeepSeek-V2.5 (Sep 2024) vs GLM-4.6

> GLM-4.6 is the stronger model overall, scoring 40.9 to 37.0 on the Noometry Index.

- Canonical page: https://noometry.com/compare/deepseek-v2-5-vs-glm-4-6
- Last updated: 2026-10-10
- Shared benchmarks: 17

## Summary

- They share 17 benchmarks with published results for both. DeepSeek-V2.5 (Sep 2024) scores higher in 1 category and GLM-4.6 in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where GLM-4.6 leads 61.3 to 50.0.

## Snapshot

| | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| Provider | DeepSeek | Z.ai (Zhipu) |
| Noometry Index | 37.0 | 40.9 |
| Rank | 197 | 132 |
| Context | — | 205K |
| Input $/M | — | $0.60 |
| Output $/M | — | $2.20 |
| Weights | Open | Open |

## Coding

- DeepSeek-V2.5 (Sep 2024): 31.3 (#274)
- GLM-4.6: 39.9 (#137)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| LMArena Coding | 1310 | 1449 |
| SWE-bench Verified (bash only) | — | 55.4% |
| Aider Polyglot | 17.8% | — |
| LMArena WebDev | — | 1339 |
| SciCode | — | 38.4% |
| BigCodeBench Instruct | 48.6% | — |
| BigCodeBench Complete | 53.2% | — |
| ALE-Bench | — | 340.82 |
| HumanEval+ | 83.5% | — |
| MBPP+ | 74.1% | — |

## Agentic & Tool Use

- DeepSeek-V2.5 (Sep 2024): —
- GLM-4.6: 32.1 (#60)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| Terminal-Bench | — | 24.5% |
| Berkeley Function Calling Leaderboard | — | 72.4% |

## Reasoning

- DeepSeek-V2.5 (Sep 2024): 24.5 (#145)
- GLM-4.6: 23.1 (#168)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| LMArena Hard Prompts | 1289 | 1440 |
| Kagi LLM Benchmark | — | 47.4% |
| CritPt | — | 1.1% |

## Math

- DeepSeek-V2.5 (Sep 2024): 33.3 (#177)
- GLM-4.6: 36.4 (#115)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| LMArena Math | 1289 | 1431 |
| FrontierMath (Feb 2025 set) | — | 3.8% |
| FrontierMath Tier 4 (v1) | — | 2.1% |

## Knowledge

- DeepSeek-V2.5 (Sep 2024): 34.2 (#190)
- GLM-4.6: 39.6 (#118)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| LMArena Expert | 1266 | 1432 |
| Vectara Hallucination Rate | — | 9.5% |

## Multilingual

- DeepSeek-V2.5 (Sep 2024): 42.7 (#188)
- GLM-4.6: 53.6 (#62)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| LMArena Non-English | 1273 | 1426 |
| LMArena Chinese | 1317 | 1500 |
| LMArena French | 1288 | 1459 |
| LMArena German | 1258 | 1447 |
| LMArena Japanese | 1228 | 1394 |
| LMArena Korean | 1209 | 1401 |
| LMArena Russian | 1289 | 1418 |
| LMArena Spanish | 1248 | 1436 |

## Instruction Following

- DeepSeek-V2.5 (Sep 2024): 67.7 (#189)
- GLM-4.6: 74.5 (#93)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| LMArena Instruction Following | 1280 | 1410 |

## Long Context

- DeepSeek-V2.5 (Sep 2024): 39.5 (#169)
- GLM-4.6: 43.4 (#91)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| LMArena Longer Query | 1301 | 1423 |

## Writing & Preference

- DeepSeek-V2.5 (Sep 2024): 50.0 (#182)
- GLM-4.6: 61.3 (#86)

| Benchmark | DeepSeek-V2.5 (Sep 2024) | GLM-4.6 |
|---|---|---|
| LMArena Text | 1294 | 1440 |
| LMArena Creative Writing | 1285 | 1411 |
| LMArena Multi-Turn | 1297 | 1427 |
| EQ-Bench Creative Writing | — | 1411 |

## FAQ

### Is DeepSeek-V2.5 (Sep 2024) better than GLM-4.6?

GLM-4.6 is the stronger model overall, scoring 40.9 to 37.0 on the Noometry Index.

### Is DeepSeek-V2.5 (Sep 2024) or GLM-4.6 better for coding?

GLM-4.6 scores higher on coding benchmarks: 39.9 versus 31.3 in the Noometry coding category.

### How many benchmarks do DeepSeek-V2.5 (Sep 2024) and GLM-4.6 share?

17 benchmarks have published results for both models. DeepSeek-V2.5 (Sep 2024) has 22 scored results on Noometry and GLM-4.6 has 29.