# DeepSeek-V3 vs GLM-5.3

> GLM-5.3 is the stronger model overall, scoring 54.8 to 39.5 on the Noometry Index. DeepSeek-V3 costs 5.3× less per token, which makes it the better buy when GLM-5.3's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/deepseek-v3-vs-glm-5-3
- Last updated: 2026-10-10
- Shared benchmarks: 26

## Summary

- They share 26 benchmarks with published results for both. DeepSeek-V3 scores higher in 0 categories and GLM-5.3 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GLM-5.3 leads 62.3 to 32.1.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 91.1% for GLM-5.3.
- DeepSeek-V3 is cheaper at $0.24 / $0.90 per million input/output tokens, against $1.40 / $4.40 for GLM-5.3.
- GLM-5.3 accepts more context: 1M tokens versus 164K.

## Snapshot

| | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| Provider | DeepSeek | Z.ai (Zhipu) |
| Noometry Index | 39.5 | 54.8 |
| Rank | 166 | 26 |
| Context | 164K | 1M |
| Input $/M | $0.24 | $1.40 |
| Output $/M | $0.90 | $4.40 |
| Weights | Open | Open |

## Coding

- DeepSeek-V3: 42.3 (#106)
- GLM-5.3: 59.5 (#14)

| Benchmark | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| SciCode | 35.8% | 59% |
| WeirdML | 36.1% | 75.4% |
| LMArena Coding | 1368 | 1496 |
| DeepSWE | — | 69% |
| FrontierCode | — | 40.1% |
| Aider Polyglot | 55.1% | — |
| CursorBench | — | 42.6% |
| LMArena WebDev | — | 1622 |
| FrontierSWE | — | 30.2% |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| ALE-Bench | — | 1,317 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |

## Agentic & Tool Use

- DeepSeek-V3: —
- GLM-5.3: 36.4 (#38)

| Benchmark | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| APEX-Agents | — | 56.6% |
| METR Time Horizons | 49.6% | — |
| Vending-Bench 2 | — | 8,164 |

## Reasoning

- DeepSeek-V3: 20.5 (#236)
- GLM-5.3: 46.1 (#46)

| Benchmark | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| CritPt | 0% | 19.1% |
| LMArena Hard Prompts | 1365 | 1489 |
| DTBench | 64.8% | 87.7% |
| LMCA | 15.5% | 55.5% |
| Epoch Capabilities Index | 135.94 | 155.61 |
| SimpleBench | 27.2% | — |
| Kagi LLM Benchmark | 52.3% | — |
| NYT Connections (extended) | — | 74.2% |
| Chess Puzzles | — | 21% |
| LiveBench Reasoning | 65.8% | — |
| Mystery Game Puzzles | — | 33% |
| LiveBench Data Analysis | 60.9% | — |
| Bench to the Future 3 | — | 0.15 |
| BIG-Bench Hard | 87.5% | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |

## Math

- DeepSeek-V3: 32.1 (#219)
- GLM-5.3: 62.3 (#33)

| Benchmark | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 91.1% |
| LMArena Math | 1373 | 1489 |
| FrontierMath (Tiers 1-3) | — | 68.8% |
| FrontierMath Tier 4 | — | 29.3% |
| ProofBench | — | 49% |
| Omni-MATH | 40.3% | — |
| LiveBench Math | 73.5% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |

## Knowledge

- DeepSeek-V3: 37.5 (#155)
- GLM-5.3: 58.3 (#37)

| Benchmark | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| GPQA Diamond | 67.6% | 90.9% |
| LMArena Expert | 1351 | 1516 |
| SimpleQA Verified | — | 41% |
| MMLU-Pro | 72.3% | — |
| Confabulations | 26.1% | — |
| Vectara Hallucination Rate | 6.1% | — |
| GPQA (HELM) | 53.8% | — |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |

## Multilingual

- DeepSeek-V3: 48.5 (#143)
- GLM-5.3: 55.7 (#28)

| Benchmark | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| LMArena Non-English | 1358 | 1457 |
| LMArena Chinese | 1391 | 1528 |
| LMArena French | 1385 | 1499 |
| LMArena German | 1374 | 1499 |
| LMArena Japanese | 1333 | 1453 |
| LMArena Korean | 1319 | 1472 |
| LMArena Russian | 1373 | 1463 |
| LMArena Spanish | 1358 | 1460 |

## Instruction Following

- DeepSeek-V3: 72.8 (#130)
- GLM-5.3: 77.5 (#23)

| Benchmark | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| LMArena Instruction Following | 1345 | 1477 |
| LiveBench Instruction Following | 81.5% | — |
| IFEval | 83.2% | — |

## Long Context

- DeepSeek-V3: 34.0 (#253)
- GLM-5.3: 45.4 (#41)

| Benchmark | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| LMArena Longer Query | 1352 | 1482 |
| Fiction.LiveBench | 50% | — |

## Writing & Preference

- DeepSeek-V3: 57.4 (#130)
- GLM-5.3: 75.7 (#6)

| Benchmark | DeepSeek-V3 | GLM-5.3 |
|---|---|---|
| LMArena Text | 1375 | 1471 |
| LMArena Creative Writing | 1364 | 1457 |
| EQ-Bench Creative Writing | 1472 | 2075 |
| LMArena Multi-Turn | 1389 | 1472 |
| Short-Story Creative Writing | 77% | — |
| WildBench | 83% | — |
| LiveBench Language | 49.1% | — |

## FAQ

### Is DeepSeek-V3 better than GLM-5.3?

GLM-5.3 is the stronger model overall, scoring 54.8 to 39.5 on the Noometry Index. DeepSeek-V3 costs 5.3× less per token, which makes it the better buy when GLM-5.3's lead doesn't matter for your workload.

### Which is cheaper, DeepSeek-V3 or GLM-5.3?

DeepSeek-V3 is cheaper. It lists at $0.24 per million input tokens and $0.90 per million output tokens; GLM-5.3 lists at $1.40 and $4.40.

### Is DeepSeek-V3 or GLM-5.3 better for coding?

GLM-5.3 scores higher on coding benchmarks: 59.5 versus 42.3 in the Noometry coding category.

### Which has the bigger context window?

GLM-5.3 does, with 1M tokens against 164K.

### How many benchmarks do DeepSeek-V3 and GLM-5.3 share?

26 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and GLM-5.3 has 42.