# DeepSeek-R1 vs GLM-5.2

> GLM-5.2 is the stronger model overall, scoring 50.9 to 42.0 on the Noometry Index. DeepSeek-R1 costs 2.4× less per token, which makes it the better buy when GLM-5.2's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/deepseek-r1-vs-glm-5-2
- Last updated: 2026-10-10
- Shared benchmarks: 29

## Summary

- They share 29 benchmarks with published results for both. DeepSeek-R1 scores higher in 1 category and GLM-5.2 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GLM-5.2 leads 42.1 to 18.7.
- The biggest single-benchmark swing is ARC-AGI-1: 21.2% for DeepSeek-R1 and 77% for GLM-5.2.
- DeepSeek-R1 is cheaper at $0.50 / $2.15 per million input/output tokens, against $1.40 / $4.40 for GLM-5.2.
- GLM-5.2 accepts more context: 1M tokens versus 164K.
- GLM-5.2 has downloadable open weights; the other is API-only.

## Snapshot

| | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| Provider | DeepSeek | Z.ai (Zhipu) |
| Noometry Index | 42.0 | 50.9 |
| Rank | 110 | 38 |
| Context | 164K | 1M |
| Input $/M | $0.50 | $1.40 |
| Output $/M | $2.15 | $4.40 |
| Weights | Proprietary | Open |

## Coding

- DeepSeek-R1: 45.9 (#63)
- GLM-5.2: 50.9 (#37)

| Benchmark | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| SciCode | 35.7% | 50.5% |
| WeirdML | 41.6% | 70.1% |
| LMArena Coding | 1427 | 1484 |
| ALE-Bench | 804.12 | 1,047 |
| SWE-bench Verified | — | 78.7% |
| DeepSWE | — | 43.8% |
| FrontierCode | — | 24.5% |
| Aider Polyglot | 71.4% | — |
| LMArena WebDev | — | 1604 |
| LiveBench Coding | 66.7% | — |
| AlgoTune | 1.7 | — |

## Agentic & Tool Use

- DeepSeek-R1: 30.4 (#68)
- GLM-5.2: 31.6 (#63)

| Benchmark | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| τ²-bench Banking | — | 37.1% |
| DeepResearch Bench | 35.1% | — |
| PostTrainBench | — | 31.7% |
| BALROG | 34.9% | — |
| GBAEval | — | 0% |
| METR Time Horizons | 53.8% | — |
| Vending-Bench 2 | — | 8,314 |

## Reasoning

- DeepSeek-R1: 18.7 (#253)
- GLM-5.2: 42.1 (#46)

| Benchmark | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| ARC-AGI-2 | 1.3% | 22.8% |
| SimpleBench | 40.8% | 58.8% |
| Kagi LLM Benchmark | 69.4% | 62.6% |
| ARC-AGI-1 | 21.2% | 77% |
| CritPt | 1.1% | 20.9% |
| LMArena Hard Prompts | 1415 | 1480 |
| Epoch Capabilities Index | 141.3 | 151.76 |
| NYT Connections (extended) | — | 74.3% |
| Chess Puzzles | — | 21% |
| EBR-Bench | — | 9.5% |
| LiveBench Reasoning | 83.2% | — |
| Mystery Game Puzzles | — | 19% |
| DTBench | — | 93.6% |
| LiveBench Data Analysis | 69.8% | — |
| LMCA | — | 45.8% |
| Surface Evolver Bench | — | 55.6% |
| ForecastBench | 60 | — |
| LiveBench | 71.6% | — |

## Math

- DeepSeek-R1: 42.0 (#70)
- GLM-5.2: 55.4 (#35)

| Benchmark | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.4% | 86.4% |
| LMArena Math | 1401 | 1482 |
| FrontierMath (Tiers 1-3) | — | 59.2% |
| FrontierMath Tier 4 | — | 29.3% |
| MathArena Final-Answer Competitions | — | 67.6% |
| ProofBench | — | 35% |
| Omni-MATH | 42.4% | — |
| LiveBench Math | 80.7% | — |
| MATH Level 5 | 96.6% | — |

## Knowledge

- DeepSeek-R1: 44.3 (#81)
- GLM-5.2: 57.1 (#36)

| Benchmark | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| GPQA Diamond | 76.3% | 91.9% |
| LMArena Expert | 1394 | 1484 |
| SimpleQA Verified | — | 34.2% |
| MMLU-Pro | 79.3% | — |
| Confabulations | 12.7% | — |
| Vectara Hallucination Rate | 11.3% | — |
| GPQA (HELM) | 66.6% | — |

## Multilingual

- DeepSeek-R1: 52.6 (#80)
- GLM-5.2: 56.0 (#25)

| Benchmark | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| LMArena Non-English | 1412 | 1459 |
| LMArena Chinese | 1443 | 1518 |
| LMArena French | 1415 | 1476 |
| LMArena German | 1405 | 1470 |
| LMArena Japanese | 1394 | 1451 |
| LMArena Korean | 1359 | 1445 |
| LMArena Russian | 1423 | 1467 |
| LMArena Spanish | 1410 | 1477 |

## Instruction Following

- DeepSeek-R1: 72.2 (#138)
- GLM-5.2: 77.1 (#32)

| Benchmark | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| LMArena Instruction Following | 1382 | 1465 |
| LiveBench Instruction Following | 80.5% | — |
| IFEval | 78.4% | — |

## Long Context

- DeepSeek-R1: 45.4 (#36)
- GLM-5.2: 45.3 (#43)

| Benchmark | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| LMArena Longer Query | 1391 | 1479 |
| Fiction.LiveBench | 75% | — |

## Writing & Preference

- DeepSeek-R1: 61.5 (#84)
- GLM-5.2: 70.7 (#20)

| Benchmark | DeepSeek-R1 | GLM-5.2 |
|---|---|---|
| LMArena Text | 1428 | 1470 |
| LMArena Creative Writing | 1406 | 1465 |
| EQ-Bench Creative Writing | 1500 | 1757 |
| LMArena Multi-Turn | 1404 | 1468 |
| Short-Story Creative Writing | 83% | — |
| WildBench | 82.8% | — |
| EQ-Bench 4 | — | 1222 |
| LiveBench Language | 48.5% | — |

## FAQ

### Is DeepSeek-R1 better than GLM-5.2?

GLM-5.2 is the stronger model overall, scoring 50.9 to 42.0 on the Noometry Index. DeepSeek-R1 costs 2.4× less per token, which makes it the better buy when GLM-5.2's lead doesn't matter for your workload.

### Which is cheaper, DeepSeek-R1 or GLM-5.2?

DeepSeek-R1 is cheaper. It lists at $0.50 per million input tokens and $2.15 per million output tokens; GLM-5.2 lists at $1.40 and $4.40.

### Is DeepSeek-R1 or GLM-5.2 better for coding?

GLM-5.2 scores higher on coding benchmarks: 50.9 versus 45.9 in the Noometry coding category.

### Which has the bigger context window?

GLM-5.2 does, with 1M tokens against 164K.

### How many benchmarks do DeepSeek-R1 and GLM-5.2 share?

29 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and GLM-5.2 has 50.