Compare / head-to-head
GLM-4.5VvsGLM-4.6V
GLM-4.6V leads 9 of 11 shared benchmarks. GLM-4.6V is 2x cheaper per token. GLM-4.6V has the larger context window (128K tokens).
Benchmarks from cited public sources; pricing from official pages; status from official provider feeds.
Shared benchmarks
0 – 9
GLM-4.6V leads · 2 tied
Cheaper per token
GLM-4.6V
2x cheaper, input + output
Larger context
GLM-4.6V
128K tokens
Providers
Z.ai
same provider
GLM-4.5V
Z.ai · released 2025-08-11
textvisionvideo
Quality
Benchmark matrix
11 shared · 6 only GLM-4.5V · 6 only GLM-4.6V| Benchmark | GLM-4.5V | GLM-4.6V | Δ | Edge |
|---|---|---|---|---|
| Reported by both · 11 | ||||
| LMArena Elo | 1352.4 ↗ | 1378.7 ↗ | -26.3 | GLM-4.6V |
| AndroidWorld | 57% ↗ | 57% ↗ | 0 | Tie |
| ChartQAPro | 64% ↗ | 65.5% ↗ | -1.5 pt | GLM-4.6V |
| Design2Code | 82.2% ↗ | 88.6% ↗ | -6.4 pt | GLM-4.6V |
| LMArena Vision (LMArena) | 1152.7 ↗ | 1161.5 ↗ | -8.8 | GLM-4.6V |
| MathVista | 84.6% ↗ | 85.2% ↗ | -0.6 pt | GLM-4.6V |
| MMLongBench-Doc | 44.7% ↗ | 54.9% ↗ | -10.2 pt | GLM-4.6V |
| MMStar | 75.3% ↗ | 75.9% ↗ | -0.6 pt | GLM-4.6V |
| OCRBench | 86.5% ↗ | 86.5% ↗ | 0 | Tie |
| OSWorld | 35.8% ↗ | 37.2% ↗ | -1.4 pt | GLM-4.6V |
| VideoMMMU | 72.4% ↗ | 74.7% ↗ | -2.3 pt | GLM-4.6V |
| Only GLM-4.5V reports · 6 | ||||
| MathVision | 65.6% ↗ | not reported | — | — |
| MMBench v1.1 | 88.2% ↗ | not reported | — | — |
| MMMU (val) | 75.4% ↗ | not reported | — | — |
| MMMU Pro | 65.2% ↗ | not reported | — | — |
| VideoMME (w/o sub) | 74.6% ↗ | not reported | — | — |
| WebVoyagerSom | 84.4% ↗ | not reported | — | — |
| Only GLM-4.6V reports · 6 | ||||
| CharXiv_Val-Reasoning | not reported | 63.2% ↗ | — | — |
| MMBench V1.1 | not reported | 88.8% ↗ | — | — |
| MMBrowseComp | not reported | 7.6% ↗ | — | — |
| MMMU (Val) | not reported | 76% ↗ | — | — |
| MMMU_Pro | not reported | 66% ↗ | — | — |
| WebVoyager | not reported | 81% ↗ | — | — |
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-4.5V minus GLM-4.6V 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-4.5V | GLM-4.6V | Edge |
|---|---|---|---|
| Context window | 64K tokens | 128K tokens | GLM-4.6V |
| Max output | 16K tokens | 32K tokens | GLM-4.6V |
| Input price / 1M | $0.6 ↗ | $0.3 ↗ | GLM-4.6V |
| Output price / 1M | $1.8 ↗ | $0.9 ↗ | GLM-4.6V |
| Cached input / 1M | $0.11 ↗ | $0.05 ↗ | GLM-4.6V |
| Input + output / 1M Lower is cheaper. List prices; batch, tool and regional fees excluded. | $2.4 | $1.2 | GLM-4.6V |
| Modalities | text · vision · video | text · vision · video | Tie |
| Released | 2025-08-11 | 2025-12-08 | — |
| Cited benchmark scores | 17 | 17 | — |
Reliability
Z.ai status
Live from /statusMore matchups
GLM-4.5V vs …
Models sharing the most benchmarksMore matchups
GLM-4.6V vs …
Models sharing the most benchmarksBuilt by Respan
Which one wins on your data?
Public benchmarks are a starting point. Run GLM-4.5V and GLM-4.6V on your own prompts with Respan evals, or route to either through one gateway key with automatic failover.