Model comparison
DeepSeek-R1 vs GLM-4.5-Air
DeepSeek-R1 is the stronger model overall, scoring 42.0 to 38.7 on the Noometry Index. GLM-4.5-Air costs 2.1× less per token, which makes it the better buy when DeepSeek-R1's lead doesn't matter for your workload.
Last verified . 25 shared benchmarks.
DeepSeekSummary
- They share 25 benchmarks with published results for both. DeepSeek-R1 scores higher in 7 categories and GLM-4.5-Air in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in coding, where DeepSeek-R1 leads 45.9 to 34.5.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 69.4% for DeepSeek-R1 and 43% for GLM-4.5-Air.
- GLM-4.5-Air is cheaper at $0.20 / $1.10 per million input/output tokens, against $0.50 / $2.15 for DeepSeek-R1.
- DeepSeek-R1 accepts more context: 164K tokens versus 131K.
- GLM-4.5-Air has downloadable open weights; the other is API-only.
Side by side
| DeepSeek-R1 | GLM-4.5-Air | |
|---|---|---|
| Provider | DeepSeek | Z.ai (Zhipu) |
| Noometry Index | 42.0 | 38.7 |
| Released | 2025-01-20 | 2025-07-20 |
| Weights | Proprietary | Open |
| Context window | 164K | 131K |
| Max output | 64K | 98K |
| Input $ / M tokens | $0.50 | $0.20 |
| Output $ / M tokens | $2.15 | $1.10 |
| Results tracked | 52 | 27 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-R1 leads
DeepSeek-R1: 45.9 (#63), GLM-4.5-Air: 34.5 (#225)
| Benchmark | DeepSeek-R1 | GLM-4.5-Air |
|---|---|---|
| LMArena Coding | 1427 | 1397 |
| Aider Polyglot | 71.4% | — |
| SciCode | 35.7% | — |
| GSO | — | 2.9% |
| WeirdML | 41.6% | — |
| LiveBench Coding | 66.7% | — |
| ALE-Bench | 804.12 | — |
| AlgoTune | 1.7 | — |
Agentic & Tool Use Not comparable
DeepSeek-R1: 30.4 (#68), GLM-4.5-Air: —
| Benchmark | DeepSeek-R1 | GLM-4.5-Air |
|---|---|---|
| DeepResearch Bench | 35.1% | — |
| BALROG | 34.9% | — |
| METR Time Horizons | 53.8% | — |
Reasoning GLM-4.5-Air leads
DeepSeek-R1: 18.7 (#253), GLM-4.5-Air: 23.3 (#166)
| Benchmark | DeepSeek-R1 | GLM-4.5-Air |
|---|---|---|
| Kagi LLM Benchmark | 69.4% | 43% |
| LMArena Hard Prompts | 1415 | 1379 |
| ForecastBench | 60 | 59.2 |
| ARC-AGI-2 | 1.3% | — |
| SimpleBench | 40.8% | — |
| ARC-AGI-1 | 21.2% | — |
| CritPt | 1.1% | — |
| LiveBench Reasoning | 83.2% | — |
| LiveBench Data Analysis | 69.8% | — |
| Epoch Capabilities Index | 141.3 | — |
| LiveBench | 71.6% | — |
Math DeepSeek-R1 leads
DeepSeek-R1: 42.0 (#70), GLM-4.5-Air: 33.8 (#166)
| Benchmark | DeepSeek-R1 | GLM-4.5-Air |
|---|---|---|
| Omni-MATH | 42.4% | 39.1% |
| LMArena Math | 1401 | 1396 |
| OTIS Mock AIME 2024-2025 | 66.4% | — |
| LiveBench Math | 80.7% | — |
| MATH Level 5 | 96.6% | — |
Knowledge DeepSeek-R1 leads
DeepSeek-R1: 44.3 (#81), GLM-4.5-Air: 34.6 (#184)
| Benchmark | DeepSeek-R1 | GLM-4.5-Air |
|---|---|---|
| MMLU-Pro | 79.3% | 76.2% |
| Vectara Hallucination Rate | 11.3% | 9.3% |
| GPQA (HELM) | 66.6% | 59.4% |
| LMArena Expert | 1394 | 1370 |
| GPQA Diamond | 76.3% | — |
| Humanity's Last Exam | — | 8.1% |
| Confabulations | 12.7% | — |
Multilingual DeepSeek-R1 leads
DeepSeek-R1: 52.6 (#80), GLM-4.5-Air: 49.4 (#130)
| Benchmark | DeepSeek-R1 | GLM-4.5-Air |
|---|---|---|
| LMArena Non-English | 1412 | 1366 |
| LMArena Chinese | 1443 | 1427 |
| LMArena French | 1415 | 1399 |
| LMArena German | 1405 | 1376 |
| LMArena Japanese | 1394 | 1348 |
| LMArena Korean | 1359 | 1306 |
| LMArena Russian | 1423 | 1373 |
| LMArena Spanish | 1410 | 1385 |
Instruction Following DeepSeek-R1 leads
DeepSeek-R1: 72.2 (#138), GLM-4.5-Air: 69.8 (#165)
| Benchmark | DeepSeek-R1 | GLM-4.5-Air |
|---|---|---|
| IFEval | 78.4% | 81.2% |
| LMArena Instruction Following | 1382 | 1353 |
| LiveBench Instruction Following | 80.5% | — |
Long Context DeepSeek-R1 leads
DeepSeek-R1: 45.4 (#36), GLM-4.5-Air: 41.6 (#130)
| Benchmark | DeepSeek-R1 | GLM-4.5-Air |
|---|---|---|
| LMArena Longer Query | 1391 | 1367 |
| Fiction.LiveBench | 75% | — |
Writing & Preference DeepSeek-R1 leads
DeepSeek-R1: 61.5 (#84), GLM-4.5-Air: 56.1 (#134)
| Benchmark | DeepSeek-R1 | GLM-4.5-Air |
|---|---|---|
| LMArena Text | 1428 | 1384 |
| LMArena Creative Writing | 1406 | 1343 |
| WildBench | 82.8% | 78.9% |
| LMArena Multi-Turn | 1404 | 1371 |
| Short-Story Creative Writing | 83% | — |
| EQ-Bench Creative Writing | 1500 | — |
| LiveBench Language | 48.5% | — |
Frequently asked questions
Is DeepSeek-R1 better than GLM-4.5-Air?
DeepSeek-R1 is the stronger model overall, scoring 42.0 to 38.7 on the Noometry Index. GLM-4.5-Air costs 2.1× less per token, which makes it the better buy when DeepSeek-R1's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-R1 or GLM-4.5-Air?
GLM-4.5-Air is cheaper. It lists at $0.20 per million input tokens and $1.10 per million output tokens; DeepSeek-R1 lists at $0.50 and $2.15.
Is DeepSeek-R1 or GLM-4.5-Air better for coding?
DeepSeek-R1 scores higher on coding benchmarks: 45.9 versus 34.5 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-R1 does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-R1 and GLM-4.5-Air share?
25 benchmarks have published results for both models. DeepSeek-R1 has 52 scored results on Noometry and GLM-4.5-Air has 27.