Compare / GPT-5.4 vs GLM 5.2
GPT-5.4 vs GLM 5.2
28 results from 6 sources that measured both models. Each row is in the source's own units; there is no overall winner.
| Benchmark | Metric | GPT-5.4 | GLM 5.2 |
|---|---|---|---|
| Arena (formerly LMArena)reported | Confirmed task successIPS effect estimate | -0.04 | 0.04 |
| Arena (formerly LMArena)reported | Praise over complaintIPS effect estimate | -0.01 | 0.10 |
| Arena (formerly LMArena)reported | SteerabilityIPS effect estimate | 0.02 | 0.04 |
| Arena (formerly LMArena)reported | Tool groundingIPS effect estimate | 0.00 | 0.00 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,485 | 1,463 |
| Arena (formerly LMArena)reported | Business, management and financeArena rating | 1,482 | 1,463 |
| Arena (formerly LMArena)reported | Creative writingArena rating | 1,443 | 1,455 |
| Arena (formerly LMArena)reported | Expert promptsArena rating | 1,515 | 1,500 |
| Arena (formerly LMArena)reported | Instruction followingArena rating | 1,471 | 1,470 |
| Arena (formerly LMArena)reported | OverallArena rating | 1,475 | 1,476 |
| Arena (formerly LMArena)reported | Writing, literature and languageArena rating | 1,465 | 1,459 |
| Artificial Analysisreported | Artificial Analysis Coding Indexindex score | 71.1 | 68.8 |
| Artificial Analysisreported | Artificial Analysis Intelligence Indexindex score | 39.0 | 33.7 |
| Artificial Analysisreported | GPQA Diamond% of questions | 92.0% | 89.5% |
| Artificial Analysisreported | Humanity's Last Exam% of questions | 43.7% | 41.1% |
| Artificial Analysisreported | IFBench% of instructions | 73.9% | 73.3% |
| Artificial Analysisreported | Long-context reasoning (AA-LCR)% of questions | 82.0% | 78.3% |
| Artificial Analysisreported | Terminal-Bench 2.1% of tasks | 78.3% | 77.9% |
| Artificial Analysisreported | Terminal-Bench Hard% of tasks | 57.6% | 50.8% |
| Artificial Analysisreported | Τ-bench banking% of tasks | 39.6% | 34.6% |
| Artificial Analysisreported | Τ²-bench telecom% of tasks | 87.1% | 99.1% |
| SimpleQA Verified (Epoch AI)reported | Correct answers% of questions | 45.1% | 34.2% |
| tau2-benchreported | Consistency (pass^4)% of tasks | 21.6% | 13.4% |
| tau2-benchreported | Task success (pass^1)% of tasks | 39.4% | 37.1% |
| UGI Leaderboardreported | Requested-length error% off the requested word count | 30.0% | 12.0% |
| UGI Leaderboardreported | Style adherencescore from 0 to 1 | 0.40 | 0.37 |
| UGI Leaderboardreported | Writing scorescore out of 100 | 68.8 | 67.0 |
| Vending-Bench 2reported | Money after a yearUS dollars | $6,144.18 | $8,313.78 |
Bold green marks the better value on that metric. Where a source reports ranges that overlap, the difference may not be meaningful; see the benchmark page for ranges.