Model comparison
GLM-5.1 vs Phi-4
GLM-5.1 is the stronger model overall, scoring 47.6 to 30.9 on the Noometry Index. Phi-4 costs 25× less per token, which makes it the better buy when GLM-5.1's lead doesn't matter for your workload.
Last verified . 21 shared benchmarks.
MicrosoftSummary
- They share 21 benchmarks with published results for both. GLM-5.1 scores higher in 9 categories and Phi-4 in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where GLM-5.1 leads 49.0 to 19.6.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 93.3% for GLM-5.1 and 13.8% for Phi-4.
- Phi-4 is cheaper at $0.07 / $0.14 per million input/output tokens, against $1.40 / $4.40 for GLM-5.1.
- GLM-5.1 accepts more context: 200K tokens versus 128K.
Side by side
| GLM-5.1 | Phi-4 | |
|---|---|---|
| Provider | Z.ai (Zhipu) | Microsoft |
| Noometry Index | 47.6 | 30.9 |
| Released | 2026-04-07 | 2024-12-11 |
| Weights | Open | Open |
| Context window | 200K | 128K |
| Max output | 131K | 4K |
| Input $ / M tokens | $1.40 | $0.07 |
| Output $ / M tokens | $4.40 | $0.14 |
| Results tracked | 41 | 37 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GLM-5.1 leads
GLM-5.1: 48.5 (#52), Phi-4: 33.8 (#237)
| Benchmark | GLM-5.1 | Phi-4 |
|---|---|---|
| LMArena Coding | 1486 | 1231 |
| SWE-bench Verified | 74.2% | — |
| LMArena WebDev | 1509 | — |
| SciCode | 43.8% | — |
| WeirdML | 57.1% | — |
| BigCodeBench Instruct | — | 45.5% |
| LiveBench Coding | — | 30.7% |
| BigCodeBench Complete | — | 55.4% |
| ALE-Bench | 887.1 | — |
Agentic & Tool Use GLM-5.1 leads
GLM-5.1: 24.8 (#106), Phi-4: 22.6 (#120)
| Benchmark | GLM-5.1 | Phi-4 |
|---|---|---|
| APEX-Agents | 40.9% | — |
| Berkeley Function Calling Leaderboard | — | 28.8% |
| BALROG | — | 11.6% |
| ExploitBench | 18.1% | — |
| GBAEval | 0% | — |
| Vending-Bench 2 | 5,634 | — |
Reasoning GLM-5.1 leads
GLM-5.1: 38.2 (#57), Phi-4: 17.0 (#287)
| Benchmark | GLM-5.1 | Phi-4 |
|---|---|---|
| Chess Puzzles | 19% | 1% |
| LMArena Hard Prompts | 1472 | 1220 |
| Epoch Capabilities Index | 149.86 | 130.43 |
| SimpleBench | 55.1% | — |
| NYT Connections (extended) | 77.7% | — |
| CritPt | 4.6% | — |
| Thematic Generalization | 69.8% | — |
| LiveBench Reasoning | — | 47.8% |
| LiveBench Data Analysis | — | 45.2% |
| LiveBench | — | 41.6% |
Math GLM-5.1 leads
GLM-5.1: 49.0 (#48), Phi-4: 19.6 (#279)
| Benchmark | GLM-5.1 | Phi-4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 93.3% | 13.8% |
| LMArena Math | 1472 | 1246 |
| FrontierMath (Tiers 1-3) | 36.8% | — |
| MathArena Final-Answer Competitions | 67.1% | — |
| ProofBench | 22.2% | — |
| LiveBench Math | — | 42% |
| MATH Level 5 | — | 64.9% |
| FrontierMath (Feb 2025 set) | 33.4% | — |
| FrontierMath Tier 4 (v1) | 12.5% | — |
Knowledge GLM-5.1 leads
GLM-5.1: 55.0 (#45), Phi-4: 32.5 (#204)
| Benchmark | GLM-5.1 | Phi-4 |
|---|---|---|
| GPQA Diamond | 89.9% | 56.1% |
| LMArena Expert | 1477 | 1203 |
| SimpleQA Verified | 34% | — |
| Confabulations | — | 29.4% |
| Vectara Hallucination Rate | — | 3.7% |
| MMLU | — | 84.8% |
Multilingual GLM-5.1 leads
GLM-5.1: 55.2 (#34), Phi-4: 37.4 (#232)
| Benchmark | GLM-5.1 | Phi-4 |
|---|---|---|
| LMArena Non-English | 1448 | 1197 |
| LMArena Chinese | 1516 | 1211 |
| LMArena French | 1478 | 1223 |
| LMArena German | 1466 | 1222 |
| LMArena Japanese | 1436 | 1158 |
| LMArena Korean | 1423 | 1151 |
| LMArena Russian | 1453 | 1209 |
| LMArena Spanish | 1467 | 1234 |
Instruction Following GLM-5.1 leads
GLM-5.1: 76.5 (#41), Phi-4: 60.6 (#246)
| Benchmark | GLM-5.1 | Phi-4 |
|---|---|---|
| LMArena Instruction Following | 1451 | 1201 |
| LiveBench Instruction Following | — | 58.4% |
Long Context GLM-5.1 leads
GLM-5.1: 44.9 (#51), Phi-4: 36.9 (#221)
| Benchmark | GLM-5.1 | Phi-4 |
|---|---|---|
| LMArena Longer Query | 1467 | 1217 |
Writing & Preference GLM-5.1 leads
GLM-5.1: 67.2 (#29), Phi-4: 40.7 (#239)
| Benchmark | GLM-5.1 | Phi-4 |
|---|---|---|
| LMArena Text | 1462 | 1217 |
| LMArena Creative Writing | 1454 | 1182 |
| LMArena Multi-Turn | 1473 | 1206 |
| Short-Story Creative Writing | — | 62.6% |
| EQ-Bench Creative Writing | 1592 | — |
| LiveBench Language | — | 25.6% |
Frequently asked questions
Is GLM-5.1 better than Phi-4?
GLM-5.1 is the stronger model overall, scoring 47.6 to 30.9 on the Noometry Index. Phi-4 costs 25× less per token, which makes it the better buy when GLM-5.1's lead doesn't matter for your workload.
Which is cheaper, GLM-5.1 or Phi-4?
Phi-4 is cheaper. It lists at $0.07 per million input tokens and $0.14 per million output tokens; GLM-5.1 lists at $1.40 and $4.40.
Is GLM-5.1 or Phi-4 better for coding?
GLM-5.1 scores higher on coding benchmarks: 48.5 versus 33.8 in the Noometry coding category.
Which has the bigger context window?
GLM-5.1 does, with 200K tokens against 128K.
How many benchmarks do GLM-5.1 and Phi-4 share?
21 benchmarks have published results for both models. GLM-5.1 has 41 scored results on Noometry and Phi-4 has 37.