Compare / head-to-head
GLM-5.3vs
Muse Spark 1.2
GLM-5.3 leads 12 of 19 shared benchmarks. Muse Spark 1.2 is 1.1x cheaper per token. Muse Spark 1.2 has the larger context window (1.0M tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
12 – 7
GLM-5.3 leads
Cheaper per token
Muse Spark 1.2
1.1x cheaper, input + output
Larger context
Muse Spark 1.2
1.0M tokens
Providers
2 providers
Z.ai · Meta
GLM-5.3
Z.ai · released 2026-08-18
text
Muse Spark 1.2
Meta · released 2026-08-05
textvisionvideoaudio
Quality
Benchmark matrix
19 shared · 24 only GLM-5.3 · 4 only Muse Spark 1.2| Benchmark | GLM-5.3 | Muse Spark 1.2 | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 19 | ||||
| LMArena Elo | 1478.5 ↗ | 1493.5 ↗ | -15 | Muse Spark 1.2 |
| APEX-Agents (Mercor) | 56.6% ↗ | 36.4% ↗ | +20.2 pt | GLM-5.3 |
| DeepSWE v1.1 | 66.9% ↗ | 59.3% ↗ | +7.6 pt | GLM-5.3 |
| DeepSWE v1.1 (Datacurve) | 69% ↗ | 54.9% ↗ | +14.1 pt | GLM-5.3 |
| FrontierSWE V2 (Proximal Labs) | 30.2% ↗ | 12% ↗ | +18.2 pt | GLM-5.3 |
| LiveBench Agentic Coding (LiveBench) | 60.9% ↗ | 57.6% ↗ | +3.3 pt | GLM-5.3 |
| LiveBench Coding (LiveBench) | 79% ↗ | 77.5% ↗ | +1.5 pt | GLM-5.3 |
| LiveBench Data Analysis (LiveBench) | 70.2% ↗ | 76.5% ↗ | -6.3 pt | Muse Spark 1.2 |
| LiveBench Instruction Following (LiveBench) | 69.3% ↗ | 74.3% ↗ | -5 pt | Muse Spark 1.2 |
| LiveBench Language (LiveBench) | 79.9% ↗ | 78.6% ↗ | +1.3 pt | GLM-5.3 |
| LiveBench Mathematics (LiveBench) | 87.9% ↗ | 91.2% ↗ | -3.3 pt | Muse Spark 1.2 |
| LiveBench Reasoning (LiveBench) | 85.8% ↗ | 90% ↗ | -4.2 pt | Muse Spark 1.2 |
| LMArena Agent (LMArena) | 0.0236 ↗ | -0.0327 ↗ | +0.1 | GLM-5.3 |
| LMArena WebDev (LMArena) | 1623 ↗ | 1531.8 ↗ | +91.2 | GLM-5.3 |
| SimpleBench (SimpleBench) | 66.2% ↗ | 74.5% ↗ | -8.3 pt | Muse Spark 1.2 |
| SimpleQA Verified | 41% ↗epoch run | 60.3% ↗epoch run | -19.3 pt | Muse Spark 1.2 |
| Terminal-Bench 2.1 | 88.2% ↗ | 82.9% ↗ | +5.3 pt | GLM-5.3 |
| Terminal-Bench 4.0 (Vals AI) | 38.9% ↗ | 6.1% ↗ | +32.8 pt | GLM-5.3 |
| WeirdML (Håvard Tveit Ihle) | 75.4% ↗ | 60.3% ↗ | +15.1 pt | GLM-5.3 |
| Only GLM-5.3 reports · 24 | ||||
| GPQA Diamond | 90.9% ↗epoch run | not reported | — | — |
| Agents' Last Exam | 28.5% ↗ | not reported | — | — |
| Agents' Last Exam (ALE-CLI) | 28.5% ↗ | not reported | — | — |
| AutomationBench v1.0.6 | 48.2% ↗ | not reported | — | — |
| Chess Puzzles (Epoch AI run) | 21% ↗epoch run | not reported | — | — |
| CyberGym | 84.5% ↗ | not reported | — | — |
| ExploitBench | 54.4% ↗ | not reported | — | — |
| ExploitGym 2h | 105 ↗ | not reported | — | — |
| ExploitGym 6h | 130 ↗ | not reported | — | — |
| FrontierMath Tier 4 v2 (Epoch AI run) | 29.3% ↗epoch run | not reported | — | — |
| FrontierMath Tiers 1-3 v2 (Epoch AI run) | 68.8% ↗epoch run | not reported | — | — |
| FrontierSWE | 78.1% ↗ | not reported | — | — |
| GDPval-AA v2 | 1769 ↗ | not reported | — | — |
| HLE (with tools) | 62.5% ↗ | not reported | — | — |
| Mystery Game Puzzles (Epoch AI run) | 33% ↗epoch run | not reported | — | — |
| NL2Repo | 58% ↗ | not reported | — | — |
| OTIS Mock AIME 2024-2025 (Epoch AI run) | 91.1% ↗epoch run | not reported | — | — |
| PostTrainBench | 39.8% ↗ | not reported | — | — |
| ProgramBench (Almost Solved) | 19% ↗ | not reported | — | — |
| SWE-Marathon v1.1 | 42.5% ↗ | not reported | — | — |
| Terminal-Bench 3.0 | 28.3% ↗ | not reported | — | — |
| Toolathlon Verified | 73% ↗ | not reported | — | — |
| Vending-Bench 2 (Andon Labs) | 8163.61 ↗ | not reported | — | — |
| Z.ai Code Bench (max effort) | 34.5% ↗ | not reported | — | — |
| Only Muse Spark 1.2 reports · 4 | ||||
| LMArena Vision (LMArena) | not reported | 1292.8 ↗ | — | — |
| MCP Atlas | not reported | 90.3% ↗ | — | — |
| SAGE (Vals AI) | not reported | 47.7% ↗ | — | — |
| Toolathlon-Verified (HKUST) | not reported | 75.9% ↗ | — | — |
Scores tagged "epoch" or "matharena" are independent runs, used only where the lab has not published its own; ⚠ marks rows MathArena flags as released after the competition. Higher is better on every row. Δ is GLM-5.3 minus Muse Spark 1.2 in the benchmark's own unit. "Not reported" means the lab has not published that figure; it is not a zero. ↗ opens the source.
Specs & pricing
Side by side
Official pricing pages and model cards| Spec | GLM-5.3 | Muse Spark 1.2 | Edge |
|---|---|---|---|
| Context window | 1M tokens | 1.0M tokens | Muse Spark 1.2 |
| Max output | 131K tokens | — | — |
| Input price / 1M | $1.4 ↗ | $1.25 ↗ | Muse Spark 1.2 |
| Output price / 1M | $4.4 ↗ | $4.25 ↗ | Muse Spark 1.2 |
| Cached input / 1M | $0.26 ↗ | $0.15 ↗ | Muse Spark 1.2 |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $5.8 | $5.5 | Muse Spark 1.2 |
| Modalities | text | text · vision · video · audio | Muse Spark 1.2 |
| Released | 2026-08-18 | 2026-08-05 | — |
| Cited benchmark scores | 57 | 23 | — |
Reliability
Provider status
Live from /statusMore matchups
GLM-5.3 vs …
Models sharing the most benchmarksMore matchups
Muse Spark 1.2 vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run GLM-5.3 and Muse Spark 1.2 on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.