Model comparison
DeepSeek-V3.1 vs GLM-5.2
GLM-5.2 is the stronger model overall, scoring 51.1 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 5.1× less per token, which makes it the better buy when GLM-5.2's lead doesn't matter for your workload.
Last verified . 24 shared benchmarks.
DeepSeekSummary
- They share 24 benchmarks with published results for both. DeepSeek-V3.1 scores higher in 0 categories and GLM-5.2 in 8 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where GLM-5.2 leads 55.7 to 38.9.
- The biggest single-benchmark swing is WeirdML: 38.4% for DeepSeek-V3.1 and 70.1% for GLM-5.2.
- DeepSeek-V3.1 is cheaper at $0.25 / $0.95 per million input/output tokens, against $1.40 / $4.40 for GLM-5.2.
- GLM-5.2 accepts more context: 1M tokens versus 164K.
Side by side
| DeepSeek-V3.1 | GLM-5.2 | |
|---|---|---|
| Provider | DeepSeek | Z.ai (Zhipu) |
| Noometry Index | 42.8 | 51.1 |
| Released | 2025-08-21 | 2026-06-13 |
| Weights | Open | Open |
| Context window | 164K | 1M |
| Max output | 8K | 131K |
| Input $ / M tokens | $0.25 | $1.40 |
| Output $ / M tokens | $0.95 | $4.40 |
| Results tracked | 27 | 51 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GLM-5.2 leads
DeepSeek-V3.1: 40.3 (#144), GLM-5.2: 51.3 (#41)
| Benchmark | DeepSeek-V3.1 | GLM-5.2 |
|---|---|---|
| WeirdML | 38.4% | 70.1% |
| LMArena Coding | 1417 | 1485 |
| SWE-bench Verified | — | 78.7% |
| DeepSWE | — | 43.8% |
| FrontierCode | — | 24.5% |
| LMArena WebDev | — | 1603 |
| SciCode | — | 50.5% |
| ALE-Bench | — | 1,047 |
Agentic & Tool Use Not comparable
DeepSeek-V3.1: —, GLM-5.2: 32.4 (#63)
| Benchmark | DeepSeek-V3.1 | GLM-5.2 |
|---|---|---|
| APEX-Agents | — | 45.2% |
| τ²-bench Banking | — | 37.1% |
| PostTrainBench | — | 31.7% |
| GBAEval | — | 0% |
| Vending-Bench 2 | — | 8,314 |
Reasoning GLM-5.2 leads
DeepSeek-V3.1: 27.9 (#110), GLM-5.2: 42.3 (#52)
| Benchmark | DeepSeek-V3.1 | GLM-5.2 |
|---|---|---|
| SimpleBench | 40% | 58.8% |
| Kagi LLM Benchmark | 53.2% | 62.6% |
| LMArena Hard Prompts | 1417 | 1480 |
| DTBench | 82.7% | 93.6% |
| LMCA | 24.3% | 45.8% |
| Epoch Capabilities Index | 139.92 | 151.78 |
| ARC-AGI-2 | — | 22.8% |
| NYT Connections (extended) | — | 74.3% |
| ARC-AGI-1 | — | 77% |
| CritPt | — | 20.9% |
| Chess Puzzles | — | 21% |
| EBR-Bench | — | 9.5% |
| Mystery Game Puzzles | — | 19% |
| Surface Evolver Bench | — | 55.6% |
| ForecastBench | 58 | — |
Math GLM-5.2 leads
DeepSeek-V3.1: 38.9 (#122), GLM-5.2: 55.7 (#43)
| Benchmark | DeepSeek-V3.1 | GLM-5.2 |
|---|---|---|
| LMArena Math | 1420 | 1482 |
| FrontierMath (Tiers 1-3) | — | 59.2% |
| FrontierMath Tier 4 | — | 29.3% |
| MathArena Final-Answer Competitions | — | 67.6% |
| OTIS Mock AIME 2024-2025 | — | 86.4% |
| ProofBench | — | 35% |
Knowledge GLM-5.2 leads
DeepSeek-V3.1: 43.7 (#90), GLM-5.2: 57.1 (#40)
| Benchmark | DeepSeek-V3.1 | GLM-5.2 |
|---|---|---|
| LMArena Expert | 1405 | 1486 |
| GPQA Diamond | — | 91.9% |
| SimpleQA Verified | — | 34.2% |
| Vectara Hallucination Rate | 5.5% | — |
Multilingual GLM-5.2 leads
DeepSeek-V3.1: 51.6 (#106), GLM-5.2: 55.8 (#26)
| Benchmark | DeepSeek-V3.1 | GLM-5.2 |
|---|---|---|
| LMArena Non-English | 1400 | 1459 |
| LMArena Chinese | 1469 | 1519 |
| LMArena French | 1447 | 1479 |
| LMArena German | 1411 | 1468 |
| LMArena Japanese | 1378 | 1451 |
| LMArena Korean | 1337 | 1445 |
| LMArena Russian | 1405 | 1466 |
| LMArena Spanish | 1431 | 1477 |
Instruction Following GLM-5.2 leads
DeepSeek-V3.1: 73.9 (#110), GLM-5.2: 76.9 (#34)
| Benchmark | DeepSeek-V3.1 | GLM-5.2 |
|---|---|---|
| LMArena Instruction Following | 1400 | 1465 |
Long Context GLM-5.2 leads
DeepSeek-V3.1: 36.3 (#232), GLM-5.2: 45.3 (#43)
| Benchmark | DeepSeek-V3.1 | GLM-5.2 |
|---|---|---|
| LMArena Longer Query | 1422 | 1479 |
| Fiction.LiveBench | 52.8% | — |
Writing & Preference GLM-5.2 leads
DeepSeek-V3.1: 60.3 (#98), GLM-5.2: 70.4 (#21)
| Benchmark | DeepSeek-V3.1 | GLM-5.2 |
|---|---|---|
| LMArena Text | 1420 | 1470 |
| LMArena Creative Writing | 1401 | 1462 |
| EQ-Bench Creative Writing | 1436 | 1757 |
| LMArena Multi-Turn | 1408 | 1469 |
| EQ-Bench 4 | — | 1222 |
Frequently asked questions
Is DeepSeek-V3.1 better than GLM-5.2?
GLM-5.2 is the stronger model overall, scoring 51.1 to 42.8 on the Noometry Index. DeepSeek-V3.1 costs 5.1× 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-V3.1 or GLM-5.2?
DeepSeek-V3.1 is cheaper. It lists at $0.25 per million input tokens and $0.95 per million output tokens; GLM-5.2 lists at $1.40 and $4.40.
Is DeepSeek-V3.1 or GLM-5.2 better for coding?
GLM-5.2 scores higher on coding benchmarks: 51.3 versus 40.3 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-V3.1 and GLM-5.2 share?
24 benchmarks have published results for both models. DeepSeek-V3.1 has 27 scored results on Noometry and GLM-5.2 has 51.