Model comparison
DeepSeek-V3.2-Exp vs GLM-4.5-Air
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 38.9 on the Noometry Index.
Last verified . 19 shared benchmarks.
DeepSeekSummary
- They share 19 benchmarks with published results for both. DeepSeek-V3.2-Exp 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 knowledge, where DeepSeek-V3.2-Exp leads 51.7 to 35.0.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 52.2% for DeepSeek-V3.2-Exp and 43% for GLM-4.5-Air.
- DeepSeek-V3.2-Exp is cheaper at $0.26 / $0.38 per million input/output tokens, against $0.20 / $1.10 for GLM-4.5-Air.
- DeepSeek-V3.2-Exp accepts more context: 164K tokens versus 131K.
Side by side
| DeepSeek-V3.2-Exp | GLM-4.5-Air | |
|---|---|---|
| Provider | DeepSeek | Z.ai (Zhipu) |
| Noometry Index | 44.3 | 38.9 |
| Released | 2025-09-29 | 2025-07-20 |
| Weights | Open | Open |
| Context window | 164K | 131K |
| Max output | 66K | 98K |
| Input $ / M tokens | $0.26 | $0.20 |
| Output $ / M tokens | $0.38 | $1.10 |
| Results tracked | 49 | 27 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 46.5 (#65), GLM-4.5-Air: 33.3 (#259)
| Benchmark | DeepSeek-V3.2-Exp | GLM-4.5-Air |
|---|---|---|
| LMArena Coding | 1454 | 1397 |
| SWE-bench Verified (bash only) | 70% | — |
| Aider Polyglot | 74.2% | — |
| LMArena WebDev | 1362 | — |
| SWE-bench Multilingual | 59% | — |
| SciCode | 38.9% | — |
| GSO | — | 2.9% |
| WeirdML | 39.5% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3.2-Exp: 32.7 (#59), GLM-4.5-Air: —
| Benchmark | DeepSeek-V3.2-Exp | GLM-4.5-Air |
|---|---|---|
| Terminal-Bench | 39.6% | — |
| APEX-Agents | 21.3% | — |
| Berkeley Function Calling Leaderboard | 56.7% | — |
| TheAgentCompany | 42.9% | — |
| Vending-Bench 2 | 1,034 | — |
Reasoning GLM-4.5-Air leads
DeepSeek-V3.2-Exp: 22.1 (#208), GLM-4.5-Air: 24.1 (#166)
| Benchmark | DeepSeek-V3.2-Exp | GLM-4.5-Air |
|---|---|---|
| Kagi LLM Benchmark | 52.2% | 43% |
| LMArena Hard Prompts | 1434 | 1379 |
| ARC-AGI-2 | 4% | — |
| NYT Connections (extended) | 36.7% | — |
| ARC-AGI-1 | 57% | — |
| CritPt | 2.9% | — |
| Chess Puzzles | 14% | — |
| Thematic Generalization | 65% | — |
| DTBench | 87.7% | — |
| LMCA | 29.1% | — |
| Epoch Capabilities Index | 146.27 | — |
| ForecastBench | — | 59.2 |
Math DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 41.7 (#87), GLM-4.5-Air: 36.2 (#170)
| Benchmark | DeepSeek-V3.2-Exp | GLM-4.5-Air |
|---|---|---|
| LMArena Math | 1435 | 1396 |
| MathArena Final-Answer Competitions | 57.7% | — |
| OTIS Mock AIME 2024-2025 | 87.8% | — |
| ProofBench | 8% | — |
| Omni-MATH | — | 39.1% |
| FrontierMath (Feb 2025 set) | 22.1% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 51.7 (#66), GLM-4.5-Air: 35.0 (#191)
| Benchmark | DeepSeek-V3.2-Exp | GLM-4.5-Air |
|---|---|---|
| Vectara Hallucination Rate | 5.3% | 9.3% |
| LMArena Expert | 1436 | 1370 |
| GPQA Diamond | 83.4% | — |
| Humanity's Last Exam | — | 8.1% |
| MMLU-Pro | — | 76.2% |
| GPQA (HELM) | — | 59.4% |
Multilingual DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 52.2 (#90), GLM-4.5-Air: 49.1 (#135)
| Benchmark | DeepSeek-V3.2-Exp | GLM-4.5-Air |
|---|---|---|
| LMArena Non-English | 1409 | 1366 |
| LMArena Chinese | 1461 | 1426 |
| LMArena French | 1433 | 1399 |
| LMArena German | 1440 | 1377 |
| LMArena Japanese | 1374 | 1348 |
| LMArena Korean | 1371 | 1308 |
| LMArena Russian | 1424 | 1373 |
| LMArena Spanish | 1440 | 1386 |
Instruction Following DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 74.5 (#93), GLM-4.5-Air: 69.6 (#171)
| Benchmark | DeepSeek-V3.2-Exp | GLM-4.5-Air |
|---|---|---|
| LMArena Instruction Following | 1413 | 1354 |
| IFEval | — | 81.2% |
Long Context DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 47.6 (#16), GLM-4.5-Air: 41.6 (#135)
| Benchmark | DeepSeek-V3.2-Exp | GLM-4.5-Air |
|---|---|---|
| LMArena Longer Query | 1428 | 1366 |
| Fiction.LiveBench | 83.3% | — |
| CL-bench | 13.2% | — |
| CL-bench Life | 9.5% | — |
Writing & Preference DeepSeek-V3.2-Exp leads
DeepSeek-V3.2-Exp: 62.4 (#77), GLM-4.5-Air: 55.9 (#139)
| Benchmark | DeepSeek-V3.2-Exp | GLM-4.5-Air |
|---|---|---|
| LMArena Text | 1425 | 1384 |
| LMArena Creative Writing | 1403 | 1343 |
| LMArena Multi-Turn | 1427 | 1371 |
| EQ-Bench Creative Writing | 1515 | — |
| WildBench | — | 78.9% |
Frequently asked questions
Is DeepSeek-V3.2-Exp better than GLM-4.5-Air?
DeepSeek-V3.2-Exp is the stronger model overall, scoring 44.3 to 38.9 on the Noometry Index.
Which is cheaper, DeepSeek-V3.2-Exp or GLM-4.5-Air?
DeepSeek-V3.2-Exp is cheaper. It lists at $0.26 per million input tokens and $0.38 per million output tokens; GLM-4.5-Air lists at $0.20 and $1.10.
Is DeepSeek-V3.2-Exp or GLM-4.5-Air better for coding?
DeepSeek-V3.2-Exp scores higher on coding benchmarks: 46.5 versus 33.3 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3.2-Exp does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-V3.2-Exp and GLM-4.5-Air share?
19 benchmarks have published results for both models. DeepSeek-V3.2-Exp has 49 scored results on Noometry and GLM-4.5-Air has 27.