Zhipu AIGLM-4.6V
Best configuration ranks #395 of 610 on the BenchLeader Index at 45.2 ±5.9 (thinking reasoning effort). Released 8 Dec 2025.
- Blended price
- $0.450/M
- $0.300 in · $0.900 out
- Output speed
- 56 tok/s
- First answer
- 40 s
- first token 3.39 s
- Context
- 128k
- Overall index45
- Agents & tools46
- Knowledge51
- Instruction following34
- Multimodal32
- Long context50
- Composite43
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Default” means the publisher did not say which setting was used.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| thinkingbest | 45.2 | #395 | 56 tok/s | 40 s | $0.0004 | Agents & tools 46 · Composite 43 · Instruction following 34 · Knowledge 51 · Long context 50 · Multimodal 32 |
| default | 44.8 | #405 | 57 tok/s | 3.39 s | $0.0004 | Agents & tools 36 · Coding 55 · Composite 40 · Human preference 56 · Instruction following 32 · Knowledge 46 · Long context 34 · Multimodal 37 · Reasoning 54 |
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | thinking | default | Source |
|---|---|---|---|
| LMArena Hard Prompts | – | 1383#163 | LMArena |
| GPQA Diamond (AA)not in index | 71.9%#272 | 56.6%#379 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 9.6%#265 | 3.7%#491 | Artificial Analysis |
Coding
| Benchmark | thinking | default | Source |
|---|---|---|---|
| LMArena Coding | – | 1416#160 | LMArena |
Agents & tools
| Benchmark | thinking | default | Source |
|---|---|---|---|
| Terminal-Bench Hard | 14.4%#189 | 3.0%#298 | Artificial Analysis |
| τ²-Bench Telecom (AA)not in index | 31.6%#245 | 30.7%#249 | Artificial Analysis |
Knowledge
| Benchmark | thinking | default | Source |
|---|---|---|---|
| AA-Omniscience | -26.9#206 | -37.6#246 | Artificial Analysis |
Instruction following
| Benchmark | thinking | default | Source |
|---|---|---|---|
| IFBench | 30.1%#345 | 27.9%#357 | Artificial Analysis |
Human preference
| Benchmark | thinking | default | Source |
|---|---|---|---|
| LMArena Text | – | 1378#154 | LMArena |
Multimodal
| Benchmark | thinking | default | Source |
|---|---|---|---|
| LMArena Vision | – | 1163#92 | LMArena |
| MMMU-Pro | 48.5%#209 | 42.2%#227 | Artificial Analysis |
Long context
| Benchmark | thinking | default | Source |
|---|---|---|---|
| AA-LCR | 48.7%#260 | 17.0%#373 | Artificial Analysis |
Composite
| Benchmark | thinking | default | Source |
|---|---|---|---|
| AA Intelligence Index | 11.2#290 | 8.3#365 | Artificial Analysis |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Z.ai | 27 tok/s | 3.48 s | $0.300 | $0.900 | 131k | fp8 |
| NovitaAI | 26 tok/s | 2.40 s | $0.300 | $0.900 | 131k | bf16 |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.055 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0004 | $0.0003 | 45.3 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0041 | $0.0019 | 50.6 s |
| Code edit | 6,000 / 1,500 | $0.0032 | $0.0020 | 1.1 min |
| Agentic coding session | 60,000 / 4,000 | $0.022 | $0.011 | 1.9 min |
| Structured extraction | 2,000 / 200 | $0.0008 | $0.0004 | 43.5 s |
See also
Data as of 9 Sept 2026. Compare these configurations.