GLM 5.2 vs Muse Wonder
As of 13 Oct 2026, Muse Wonder (max) leads on quality (66.3 vs 62.5 on the BenchLeader Index), but GLM 5.2 (max) costs 3.7× less, so it is the better buy unless you need the extra quality.
GLM 5.2 
Muse Wonder Key differences
- Muse Wonder (max) leads on quality: 66.3 vs 62.5.
- GLM 5.2 (max) is stronger in agents & tools, coding, human preference, instruction following, maths.
- Muse Wonder (max) is stronger in composite, knowledge, long context, reasoning.
- GLM 5.2 (max) is 3.7× cheaper ($2.15 vs $8.00 per 1M blended).
Choose GLM 5.2 (max) if
- cost matters: $2.15 against $8.00 per million tokens blended
- the wait for the first word matters: 26.2 s against 60.8 s to first answer
- you want to run it yourself: its weights are open
Choose Muse Wonder (max) if
- knowledge is the priority (73 vs 54 on the category score)
- reasoning is the priority (88 vs 70 on the category score)
- long context is the priority (66 vs 64 on the category score)
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | GLM 5.2 (max) | Muse Wonder (max) | Leader |
|---|---|---|---|
| Reasoning | 70.0 | 87.8 | Muse Wonder |
| Coding | 62.4 | – | not comparable |
| Agents & tools | 62.9 | – | not comparable |
| Maths | 57.7 | – | not comparable |
| Knowledge | 53.9 | 73.0 | Muse Wonder |
| Instruction following | 71.6 | – | not comparable |
| Human preference | 67.0 | – | not comparable |
| Long context | 64.3 | 65.5 | Muse Wonder |
| Composite | 67.4 | 78.2 | Muse Wonder |
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.
| Measure | GLM 5.2 (max) | Muse Wonder (max) |
|---|---|---|
| Blended price per 1M tokens | $2.15 | $8.00 |
| Input / output per 1M | $1.40 / $4.40 | $4.00 / $20.00 |
| Price per task: chat reply | $0.0019 | $0.0076 |
| Price per task: summarise a 30-page report | $0.019 | $0.060 |
| Price per task: code edit | $0.015 | $0.054 |
| Price per task: agentic coding session | $0.102 | $0.320 |
| Price per task: structured extraction | $0.0037 | $0.012 |
| Cost to run a full benchmark suite | $1.47 | $4.52 |
| Output speed | 86 tok/s | 85 tok/s |
| Time to first answer | 26.2 s | 60.8 s |
| Time to a complete answer | 32.0 s | 66.7 s |
Every benchmark
| Metric | GLM 5.2 (max) | Muse Wonder (max) |
|---|---|---|
| BenchLeader Index | 62.5 | 66.3 |
| Agents & tools score | 62.9 | – |
| Coding score | 62.4 | – |
| Composite score | 67.4 | 78.2 |
| Human preference score | 67.0 | – |
| Instruction following score | 71.6 | – |
| Knowledge score | 53.9 | 73.0 |
| Long context score | 64.3 | 65.5 |
| Maths score | 57.7 | – |
| Reasoning score | 70.0 | 87.8 |
| Blended price $/M | $2.15 | $8.00 |
| Output speed | 86 tok/s | 85 tok/s |
| Time to first answer | 26.2 s | 60.8 s |
| Context window | 1M | 1.0M |
| GPQA Diamond | 91.9% | – |
| FrontierMath Tiers 1–3 | 59.2% | – |
| FrontierMath Tier 4 | 29.3% | – |
| OTIS Mock AIME | 86.4% | – |
| SWE-bench Verified (Epoch) | 78.7% | – |
| SimpleQA Verified | 34.2% | – |
| SciCode | 50.5% | – |
| WeirdML | 70.1% | – |
| ProofBench | 35.0% | – |
| LMArena Text | 1475 | – |
| LMArena Hard Prompts | 1495 | – |
| LMArena Coding | 1512 | – |
| LMArena WebDev | 1603 | – |
| LMArena Agent | 3.2 | – |
| LiveBench | – | 81.2% |
| LiveBench Reasoning | – | 90.6% |
| LiveBench Coding | – | 80.7% |
| LiveBench Agentic Coding | – | 62.7% |
| LiveBench Mathematics | – | 96.5% |
| LiveBench Data Analysis | – | 81.6% |
| LiveBench Language | – | 82.9% |
| LiveBench Instruction Following | – | 73.0% |
| AA Intelligence Index v4.3.2 | 33.7 | 52.4 |
| IFBench | 73.3% | – |
| AA-LCR | 78.3% | 80.7% |
| AA-Omniscience | 4.4 | 25.5 |
| Terminal-Bench Hard | 50.8% | – |
| GPQA Diamond (AA) | 89.5% | – |
| Humanity's Last Exam (AA) | 41.1% | 50.1% |
| SciCode (AA) | 51.2% | 60.3% |
| τ²-Bench Telecom (AA) | 99.1% | – |
| Terminal-Bench 2.1 (Vals) | 67.8% | – |
| SWE-bench (Vals) | 82.8% | – |
| CritPt | 20.9% | 28.3% |
| GDPval-AA v2.1 | 43.6% | 63.5% |
| τ³-Banking (AA) | 34.6% | – |
| ITBench SRE (AA) | 42.7% | – |
| APEX-Agents (AA) | 33.7% | – |
| Code Migration | 37.9% | – |
| Legal Research Bench | 31.3% | – |
| ProgramBench | 0.5% | – |
| SkillsBench | 45.1% | – |
| SREBench | 0.0% | – |
| Vibe Code Bench v1.1 | 64.0% | – |
| LMArena Maths | 1486 | – |
| LMArena Creative Writing | 1453 | – |
| LMArena Instruction Following | 1469 | – |
| LMArena Multi-turn | 1475 | – |
| LMArena Longer Queries | 1486 | – |
| Chess Puzzles | 21.0% | – |
| EBR-bench | 9.5% | – |
| PostTrainBench | 31.7% | – |
| DeepSWE v1.1 | 43.8% | – |
| LMCA | 45.8% | – |
| DTBench | 93.6% | – |
| ALE-Bench | 1010.2 | – |
| Terminal-Bench 4.0 (AA) | 1.0% | 52.0% |
| Terminal-Bench 2.1 (AA) | 77.9% | – |
| AutomationBench | – | 68.1% |
| GDP.pdf | – | 30.6% |
| AA-Omniscience: accuracy | 24.3% | 41.1% |
| AA-Omniscience: non-hallucination | 73.7% | 73.5% |
| AA-Briefcase v1.1 | – | 1659 |
Data as of 2026-10-13. Best configuration of each model; every score links to its source on the model pages.
GLM 5.2 vs Muse Wonder: questions
- Is GLM 5.2 better than Muse Wonder?
- Muse Wonder (max) leads on quality: 66.3 vs 62.5. The BenchLeader Index combines every independent quality benchmark; Muse Wonder (max) is ahead overall as of 2026-10-13, but check the category scores for your use.
- Which is cheaper, GLM 5.2 or Muse Wonder?
- GLM 5.2 is cheaper: $2.15 against $8.00 per million tokens, blended at three input tokens per output token.
- Which is faster, GLM 5.2 or Muse Wonder?
- GLM 5.2 streams faster: 86 against 85 output tokens per second.
- Which has the larger context window?
- Muse Wonder accepts more context: 1.0M against 1M tokens.