GLM-5.3 vs Step 5 Preview
GLM-5.3 
Step 5 Preview Verdict
- GLM-5.3 (max) and Step 5 Preview are level on quality (65.8 vs 65.0).
- GLM-5.3 (max) is stronger in agents & tools, coding, composite, human preference, maths.
- Step 5 Preview is stronger in knowledge, long context, reasoning, multimodal.
- Step 5 Preview is 1.5× cheaper ($1.43 vs $2.15 per 1M blended).
| Metric | GLM-5.3 (max) | Step 5 Preview |
|---|---|---|
| BenchLeader Index | 65.8 | 65.0 |
| Agents & tools score | 62.0 | – |
| Coding score | 64.4 | – |
| Composite score | 85.9 | 84.4 |
| Human preference score | 68.5 | – |
| Knowledge score | 59.6 | 72.0 |
| Long context score | 66.3 | 70.8 |
| Maths score | 62.7 | – |
| Reasoning score | 71.6 | 82.1 |
| Multimodal score | – | 60.8 |
| Blended price $/M | $2.15 | $1.43 |
| Output speed | 63 tok/s | – |
| Time to first answer | 34.6 s | – |
| Context window | 1M | 1M |
| GPQA Diamond | 90.9% | – |
| FrontierMath Tiers 1–3 | 68.8% | – |
| FrontierMath Tier 4 | 29.3% | – |
| OTIS Mock AIME | 91.1% | – |
| SimpleQA Verified | 41.0% | – |
| Terminal-Bench | 41.8% | – |
| SciCode | 56.5% | – |
| WeirdML | 75.4% | – |
| ProofBench | 49.0% | – |
| LMArena Text | 1483 | – |
| LMArena Hard Prompts | 1507 | – |
| LMArena Coding | 1524 | – |
| LMArena WebDev | 1614 | – |
| LMArena Agent | 3.1 | – |
| AA Intelligence Index | 44.9 | 43.6 |
| AA-LCR | 79.7% | 88.3% |
| MMMU-Pro | – | 76.4% |
| AA-Omniscience | 14.3 | 16.4 |
| GPQA Diamond (AA) | 91.7% | – |
| Humanity's Last Exam (AA) | 42.3% | 46.5% |
| SciCode (AA) | 59.0% | 58.9% |
| LiveCodeBench | 80.5% | – |
| MMLU-Pro | 86.8% | – |
| IOI | 68.4% | – |
| LegalBench | 84.8% | – |
| TaxEval | 72.4% | – |
| Terminal-Bench 2.1 (Vals) | 71.5% | – |
| SWE-bench (Vals) | 95.4% | – |
| GPQA Diamond (Vals) | 88.1% | – |
| Vals Index | 57.0 | – |
| CritPt | 19.1% | 20.9% |
| GDPval (AA) | 56.7% | 53.6% |
| τ²-Bench Banking (AA) | 50.3% | – |
| Code Migration | 44.2% | – |
| Excel Modeling Benchmark | 56.3% | – |
| Finance Agent v2 | 55.8% | – |
| Harvey's Legal Agent Benchmark | 8.3% | – |
| Legal Research Bench | 49.0% | – |
| MedCode | 42.9% | – |
| MedScribe | 88.8% | – |
| MysteryMechanism | 23.0% | – |
| ProgramBench | 1.5% | – |
| Public Benefits Bench | 68.5% | – |
| SkillsBench | 47.5% | – |
| Tax Agent Bench | 73.1% | – |
| Terminal-Bench 4.0 (Vals) | 25.3% | – |
| Terminal-Bench Science | 4.3% | – |
| Vibe Code Bench 1-100 | 20.0% | – |
| Vibe Code Bench v1.1 | 78.1% | – |
| FORTRESS | 28.2% | – |
| LMArena Maths | 1504 | – |
| LMArena Creative Writing | 1462 | – |
| LMArena Instruction Following | 1480 | – |
| LMArena Multi-turn | 1493 | – |
| LMArena Longer Queries | 1492 | – |
| Chess Puzzles | 21.0% | – |
| Mystery Game Puzzles | 33.0% | – |
| DeepSWE | 69.0% | – |
| FrontierSWE | 30.2% | – |
Data as of 2026-09-19. Best configuration of each model; every score links to its source on the model pages.
GLM-5.3 vs Step 5 Preview: questions
- Is GLM-5.3 better than Step 5 Preview?
- GLM-5.3 (max) and Step 5 Preview are level on quality (65.8 vs 65.0). The BenchLeader Index combines every independent quality benchmark; GLM-5.3 (max) is ahead overall as of 2026-09-19, but check the category scores for your use.
- Which is cheaper, GLM-5.3 or Step 5 Preview?
- Step 5 Preview is cheaper: $1.43 against $2.15 per million tokens, blended at three input tokens per output token.
- Which has the larger context window?
- Both accept 1M tokens of context.