Inference
EN
Z.AI

GLM-4.6

GLM-4.6 is a reasoning model from Z.AI. 17 benchmarks count toward its index, in 6 categories.

  • Open weights
  • Reasoning
  • Input: text
  • Released 2025-09
RankedEnough results for an index.
IndexAll categories, out of 100.41.8±5.4Median 51.0
CoverageIndex weight that has results.85%
SpeedOutput tokens per second.30/sMedian 67/s
First tokenSeconds to the first output token.1.0 sMedian 1.3 s
Input / 1MUS dollars per 1M input tokens.$0.50Median $0.60
Output / 1MUS dollars per 1M output tokens.$2.00Median $2.55
ContextMaximum tokens in one request.205KMedian 262K
EloLMArena rating and rank.1440±4 #76Median 1422

Details

From OpenRouter
API IDThe model ID on OpenRouter.
z-ai/glm-4.6
Max outputMaximum output tokens per request.
131K
Knowledge cutoffLast date of training data.
31 Mar 2025
ToolsTool calls on OpenRouter.
Yes
JSON outputOutput follows a given JSON schema.
Yes
Cache read / 1MUS dollars per 1M cached tokens.
$0.10
WeightsLink to the published weights.
Hugging Face (opens in a new tab) ↗
RetiresDate OpenRouter removes the model.
—

Where to run

APIs that serve this model.
6 providers
ProviderTypeMaker, cloud, host or router.InputOutputContext
Deep InfraHost$0.50$2.00203K
OpenRouterRouter$0.50$2.00205K
IO.NETHost$0.52$2.04200K
Novita AIHost$0.55$2.20205K
Z.AIMaker$0.60$2.20205K
Zhipu AIMaker$0.60$2.20205K

US dollars per 1M tokens, list price. From models.dev (opens in a new tab) (MIT).

Run locally

356.79B parameters
16-bit
750.3 GB
8-bit
399.0 GB
4-bit
228.1 GB
Weights
Hugging Face (opens in a new tab) ↗

Memory need with a short context. The 4-bit copy needs more than 128 GB. Compare local models

Capabilities

Index per category, out of 100.
Out of 100
AgenticMulti-step tasks with tools.3 counted48.8138 / 218
CodingCode writing and repair.4 counted46.7157 / 209
ReasoningLogic problems and puzzles.2 counted31.1205 / 227
MultimodalTasks with images and text.None counted——
KnowledgeFacts and expert knowledge.4 counted39.0198 / 268
MultilingualTasks in many languages.None counted——
InstructionTasks with strict prompt rules.1 counted23.3—
MathMath problems.3 counted44.9108 / 168

Results

17 counted
BenchmarkThe benchmark name.ResultThe score from the source.PlacePlace among results on this benchmark.
AIMEMath92.7% ±0.617 / 95
SWE-bench Verified · swebench.comCoding55.4%16 / 46
Terminal-Bench 1.0Agentic42.5% ±5.517 / 47
SWE-bench Verified · SWE-bench/experimentsCoding61.8%28 / 65
LiveCodeBenchCoding81.0% ±1.164 / 142
τ²-Bench Tool-Agent-User EvaluationAgentic76.9%73 / 149
MMLU ProKnowledge82.2% ±0.478 / 135
GPQA Diamond · Vals AIKnowledge74.5% ±2.283 / 135
FrontierMath-Tier-4-2025-07-01-PrivateMath2.1% ±2.135 / 55
Critical Physics TasksReasoning0.0%150 / 214
Terminal-Bench 2.0 · Vals AIAgentic28.1% ±4.847 / 66
FrontierMath-2025-02-28-PrivateMath3.8% ±1.150 / 69
Artificial Analysis GPQA DiamondKnowledge63.2%149 / 199
Artificial Analysis Omniscience AccuracyKnowledge21.4%124 / 164
IOI v1Coding4.3% ±2.547 / 61
Artificial Analysis Humanity's Last ExamKnowledge5.5%172 / 222
Artificial Analysis IFBenchInstruction36.7%118 / 146
Artificial Analysis Long Context ReasoningReasoning26.3%188 / 217
Vibe Code Bench v1.1Coding3.1% ±1.2115 / 123
FrontierMath-Tier-4-2025-07-01-PublicMath0.0%3 / 36
MedQAHealthcare92.2% ±0.234 / 95
FrontierMath-2025-02-28-PublicMath0.0%33 / 64
MGSMMultilingual89.7% ±1.844 / 74
LegalBenchLegal79.6% ±0.595 / 149
CorpFin v2Finance56.8% ±1.088 / 131
TaxEval v2Finance66.2% ±0.8113 / 143
Finance Agent v1.1Finance36.5% ±2.843 / 51

Sources

BenchLMCC BY-NC 4.0OpenRouterTerms restrict reuseVals AINo licence statedSWE-bench teamNo licence statedEpoch AICC BYmodels.devMIT

History

5 releases
Index42.8 → 41.8
23 Oct 202612 Nov 2026
Output / 1M$2.00 → $2.00
23 Oct 202612 Nov 2026

Questions

How much does GLM-4.6 cost?
$0.50 per 1M input tokens, $2.00 per 1M output tokens.
How good is GLM-4.6?
Its index is 41.8 out of 100. That is place 200 of 283 ranked models.
What is GLM-4.6 best at?
Agentic: 48.8 out of 100, place 138 of 218.
How fast is GLM-4.6?
30 output tokens per second, 1.0 s to the first token.
What is the context window of GLM-4.6?
205K tokens.
Is GLM-4.6 open source?
Yes. Its weights are published.
When was GLM-4.6 released?
September 2025.

Badge

GLM-4.6 on Inference360: place 200 of 283, index 41.8
HTML
<a href="https://inference360.ai/models/glm-4-6"><img src="https://inference360.ai/badges/glm-4-6.svg" alt="GLM-4.6 on Inference360: place 200 of 283, index 41.8" height="20"></a>
Markdown
[![GLM-4.6 on Inference360: place 200 of 283, index 41.8](https://inference360.ai/badges/glm-4-6.svg)](https://inference360.ai/models/glm-4-6)

Compare with

GPT-OSS 20B41.7GPT-5 nano41.9Granite 4.2 30B41.6Trinity-Large-Preview42.2GLM-4.545.7

Lower price

dots3-note Preview63.4 · FreeApodex 1.1 Mini55.7 · FreeNemotron 3 Ultra50.8 · Free

More from Z.AI

All models
GLM 4.5VUnrankedGLM-4.545.7GLM-4.5-AirUnrankedGLM 4.6VUnrankedGLM-4.750.0