z-ai Open weights Reasoning

GLM 4.6 pricing & benchmarks

Released Sep 30, 2025 · served by 5 providers · ranked #60 by intelligence in our index

Input / 1M tokens$0.430
Output / 1M tokens$1.75
Context window205K
Cached input$0.100
Intelligence index28.7
Coding index45.8
Agentic index17.7
1M in + 300K out$0.95

What GLM 4.6 is

Compared with GLM-4.5, this generation brings several key improvements: Longer context window: The context window has been expanded from 128K to 200K tokens, enabling the model to handle more complex...

Providers and prices for GLM 4.6

The same model costs different amounts depending on who serves it. Prices are per 1M tokens, cheapest first. Uptime and throughput are OpenRouter's rolling measurements, not vendor claims.

ProviderInputOutputContext QuantUptime 24hThroughput
Venicecheapest $0.430 $1.75 198K fp4 99.0%
DeepInfra $0.500 $2.00 203K fp4 99.9%
Novita $0.550 $2.20 205K bf16 97.6%
Z.AI $0.600 $2.20 203K fp4 95.3%
AtlasCloud $0.600 $2.20 203K fp8 99.5%

GLM 4.6 benchmark results

Design Arena head-to-head results, as reported through the OpenRouter model API.

CategoryArenaRankEloWin rate
godotgamedev agents #13 1180 53.2%
mobileapps agents #22 1180 48.7%
androidnative agents #26 1074 52.6%
fullstack agents #27 1079 42.3%
svg models #41 1159 52.1%
gamedev models #45 1206 54.6%
3d models #51 1185 54%
dataviz models #53 1193 52.8%
uicomponent models #54 1195 54%
codecategories models #56 1196 54.3%
website models #56 1199 54.4%

Cheaper models in the same class

Models scoring within 4 points of GLM 4.6 on the Intelligence Index, but with a lower output price.

FAQ

How much does the GLM 4.6 API cost?

$0.430 per 1M input tokens and $1.75 per 1M output tokens. A typical workload of 1M input + 300K output tokens costs about $0.95.

Is GLM 4.6 free?

No. It is a paid model starting at $0.430 per 1M input tokens, though some providers offer trial credits.

Which provider is cheapest for GLM 4.6?

Venice at $0.430 input / $1.75 output per 1M tokens (fp4 quantization).

Can I self-host GLM 4.6?

Yes — weights are published as zai-org/GLM-4.6 on Hugging Face, so you can run it on your own hardware.