Model comparison
DeepSeek V4 Pro vs GLM-5.1
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 47.8 on the Noometry Index.
Last verified . 35 shared benchmarks.
DeepSeekSummary
- They share 35 benchmarks with published results for both. DeepSeek V4 Pro scores higher in 6 categories and GLM-5.1 in 3 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where DeepSeek V4 Pro leads 56.5 to 39.1.
- The biggest single-benchmark swing is Chess Puzzles: 47% for DeepSeek V4 Pro and 19% for GLM-5.1.
- DeepSeek V4 Pro is cheaper at $0.66 / $1.98 per million input/output tokens, against $1.40 / $4.40 for GLM-5.1.
- DeepSeek V4 Pro accepts more context: 1M tokens versus 200K.
Side by side
| DeepSeek V4 Pro | GLM-5.1 | |
|---|---|---|
| Provider | DeepSeek | Z.ai (Zhipu) |
| Noometry Index | 54.3 | 47.8 |
| Released | 2026-04-24 | 2026-04-07 |
| Weights | Open | Open |
| Context window | 1M | 200K |
| Max output | 393K | 131K |
| Input $ / M tokens | $0.66 | $1.40 |
| Output $ / M tokens | $1.98 | $4.40 |
| Results tracked | 48 | 41 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek V4 Pro leads
DeepSeek V4 Pro: 52.4 (#34), GLM-5.1: 48.7 (#55)
| Benchmark | DeepSeek V4 Pro | GLM-5.1 |
|---|---|---|
| SWE-bench Verified | 77.6% | 74.2% |
| LMArena WebDev | 1582 | 1508 |
| SciCode | 51% | 43.8% |
| WeirdML | 66.2% | 57.1% |
| LMArena Coding | 1470 | 1485 |
| ALE-Bench | 1,403 | 887.1 |
| FrontierCode | 28.6% | — |
Agentic & Tool Use DeepSeek V4 Pro leads
DeepSeek V4 Pro: 32.8 (#58), GLM-5.1: 24.9 (#113)
| Benchmark | DeepSeek V4 Pro | GLM-5.1 |
|---|---|---|
| APEX-Agents | 47.3% | 40.9% |
| Vending-Bench 2 | 3,285 | 5,634 |
| ExploitBench | — | 18.1% |
| GBAEval | — | 0% |
Reasoning DeepSeek V4 Pro leads
DeepSeek V4 Pro: 56.5 (#24), GLM-5.1: 39.1 (#60)
| Benchmark | DeepSeek V4 Pro | GLM-5.1 |
|---|---|---|
| NYT Connections (extended) | 91.3% | 77.7% |
| CritPt | 18% | 4.6% |
| Chess Puzzles | 47% | 19% |
| LMArena Hard Prompts | 1461 | 1472 |
| Epoch Capabilities Index | 155.31 | 149.84 |
| ARC-AGI-2 | 61.3% | — |
| SimpleBench | — | 55.1% |
| Kagi LLM Benchmark | 53.5% | — |
| ARC-AGI-1 | 90.5% | — |
| Thematic Generalization | — | 69.8% |
| Mystery Game Puzzles | 43% | — |
| DTBench | 93.9% | — |
| LMCA | 45.5% | — |
| Surface Evolver Bench | 40% | — |
| ForecastBench | 56.1 | — |
Math DeepSeek V4 Pro leads
DeepSeek V4 Pro: 64.8 (#30), GLM-5.1: 49.7 (#60)
| Benchmark | DeepSeek V4 Pro | GLM-5.1 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 64.6% | 36.8% |
| MathArena Final-Answer Competitions | 76.6% | 67.1% |
| OTIS Mock AIME 2024-2025 | 98.6% | 93.3% |
| ProofBench | 50% | 22.2% |
| LMArena Math | 1455 | 1473 |
| FrontierMath Tier 4 | 26.8% | — |
| FrontierMath (Feb 2025 set) | — | 33.4% |
| FrontierMath Tier 4 (v1) | — | 12.5% |
Knowledge DeepSeek V4 Pro leads
DeepSeek V4 Pro: 59.5 (#31), GLM-5.1: 54.9 (#50)
| Benchmark | DeepSeek V4 Pro | GLM-5.1 |
|---|---|---|
| GPQA Diamond | 91.7% | 89.9% |
| SimpleQA Verified | 52.9% | 34% |
| LMArena Expert | 1464 | 1476 |
| Vectara Hallucination Rate | 8.6% | — |
Multilingual Too close to call
DeepSeek V4 Pro: 54.4 (#45), GLM-5.1: 55.0 (#36)
| Benchmark | DeepSeek V4 Pro | GLM-5.1 |
|---|---|---|
| LMArena Non-English | 1439 | 1447 |
| LMArena Chinese | 1486 | 1515 |
| LMArena French | 1472 | 1474 |
| LMArena German | 1458 | 1465 |
| LMArena Japanese | 1445 | 1434 |
| LMArena Korean | 1447 | 1418 |
| LMArena Russian | 1453 | 1454 |
| LMArena Spanish | 1458 | 1469 |
Instruction Following Too close to call
DeepSeek V4 Pro: 76.1 (#47), GLM-5.1: 76.3 (#42)
| Benchmark | DeepSeek V4 Pro | GLM-5.1 |
|---|---|---|
| LMArena Instruction Following | 1448 | 1451 |
Long Context Too close to call
DeepSeek V4 Pro: 45.0 (#51), GLM-5.1: 44.9 (#53)
| Benchmark | DeepSeek V4 Pro | GLM-5.1 |
|---|---|---|
| LMArena Longer Query | 1458 | 1466 |
| CL-bench Life | 13.5% | — |
Writing & Preference GLM-5.1 leads
DeepSeek V4 Pro: 65.5 (#46), GLM-5.1: 66.9 (#31)
| Benchmark | DeepSeek V4 Pro | GLM-5.1 |
|---|---|---|
| LMArena Text | 1451 | 1461 |
| LMArena Creative Writing | 1446 | 1453 |
| EQ-Bench Creative Writing | 1553 | 1592 |
| LMArena Multi-Turn | 1467 | 1472 |
| EQ-Bench 4 | 1166 | — |
Frequently asked questions
Is DeepSeek V4 Pro better than GLM-5.1?
DeepSeek V4 Pro is the stronger model overall, scoring 54.3 to 47.8 on the Noometry Index.
Which is cheaper, DeepSeek V4 Pro or GLM-5.1?
DeepSeek V4 Pro is cheaper. It lists at $0.66 per million input tokens and $1.98 per million output tokens; GLM-5.1 lists at $1.40 and $4.40.
Is DeepSeek V4 Pro or GLM-5.1 better for coding?
DeepSeek V4 Pro scores higher on coding benchmarks: 52.4 versus 48.7 in the Noometry coding category.
Which has the bigger context window?
DeepSeek V4 Pro does, with 1M tokens against 200K.
How many benchmarks do DeepSeek V4 Pro and GLM-5.1 share?
35 benchmarks have published results for both models. DeepSeek V4 Pro has 48 scored results on Noometry and GLM-5.1 has 41.