Patchwire

Z.ai · Released Jul 25, 2025

GLM 4.5 Air

GLM-4.5-Air is the lightweight variant of our latest flagship model family, also purpose-built for agent-centric applications. Like GLM-4.5, it adopts the Mixture-of-Experts (MoE) architecture but with a more compact parameter...

ReasoningOpen weights
Compare in showdown → OpenRouter ↗ All Z.ai models
Released
Jul 25 2025
Context window
131K 98K out
Input / 1M tokens
$0.13
Output / 1M tokens
$0.85

Where to run it

3 providers compared

Cheapest: Novita at $0.13 / $0.85 per 1M input / output tokens — 27% less than the priciest option.

ProviderInputOutputCache readContextMax outputUptimevs. best
NovitaBF16 $0.13 $0.85 $0.025131K98K100%24h Cheapest
SiliconFlowFP8 $0.14 $0.86 —131K118K100%24h +3%
Z.AIFP8 $0.20 $1.10 $0.03131K98K99.9%24h +37%

Value

Price vs. performance

On the value frontierNo cheaper model beats GLM 4.5 Air on GSO.
2.9%31.4%59.8%88.2% $0.10$1.00$10.00 GPT-6 Astra (OpenAI) — 79.4% · $10.00/1M in Claude Fable 5.1 (Anthropic) — 88.2% · $10.00/1M in GPT-5.6 Sol (OpenAI) — 76.5% · $4.00/1M in Claude Sonnet 5 (Anthropic) — 37.3% · $2.00/1M in Claude Fable 5 (Anthropic) — 78.4% · $10.00/1M in Claude Opus 4.8 (Anthropic) — 47.1% · $5.00/1M in GPT-5.5 (OpenAI) — 40.2% · $5.00/1M in Claude Opus 4.7 (Anthropic) — 44.1% · $5.00/1M in GPT-5.4 (OpenAI) — 31.4% · $2.50/1M in Gemini 3.1 Pro Preview (Google) — 22.6% · $2.00/1M in Claude Opus 4.6 (Anthropic) — 41.2% · $5.00/1M in Gemini 3 Flash Preview (Google) — 9.8% · $0.50/1M in GPT-5.2 (OpenAI) — 27.5% · $1.75/1M in Claude Opus 4.5 (Anthropic) — 26.5% · $5.00/1M in GPT-5.1 (OpenAI) — 13.7% · $1.25/1M in Claude Sonnet 4.5 (Anthropic) — 14.7% · $3.00/1M in GPT-5 (OpenAI) — 6.9% · $1.25/1M in Gemini 2.5 Pro (Google) — 3.9% · $1.25/1M in Gemini 2.5 Pro Preview 06-05 (Google) — 3.9% · $1.25/1M in GLM 4.5 Air (Z.ai) — 2.9% · $0.13/1M in GLM 4.5 Air

GLM 4.5 Air Other Z.ai Other labs Value frontier · GSO vs. input price per 1M tokens (log scale) · Drag to pan, pinch or Ctrl + scroll to zoom, double-click to zoom in.

Benchmarks

2 benchmark scores

BenchmarkScoreRankLeader
GSOSoftware engineering · Optimize real code to make it significantly faster 2.9% #20of 20 Claude Fable 5.188.2%
Humanity's Last ExamReasoning & knowledge · Expert questions across 100+ subjects 8.1% #23of 23 GPT-6 Astra54.8%

Specs

Details

Model ID
z-ai/glm-4.5-air
Developer
Z.ai
Release date
Jul 25, 2025 ·
Context window
131,072 tokens
Max output
98,304 tokens
API price
$0.13 / $0.85 per 1M input / output tokens
Providers
3
Input types
text
Reasoning
Yes
Tool calling
Yes
Open weights
Yes
Free tier
No

Release date, prices and providers from OpenRouter's catalog; buying directly from a provider can cost a little differently. Uptime is the share of OpenRouter requests each provider served over the last day. Benchmark scores are the best result per benchmark from Epoch AI, licensed CC BY 4.0. Rank counts every model tracked here with a score.