GLM 4.5 Air
Z.ai (Zhipu AI)huggingface.co/zai-org/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...
Updated 16 min ago · first seen 11 Sept 2026
model_01M294WWE3GESV86WKRF2CMT0A
- Parameters
- 110.5B
- T2 · 2 h ago
- Context
- 128K tokens
- T2 · 27 min ago
- Released
- 25 Jul 2025
- T2 · 32 min ago
- License
- mit
- T2 · 1 h ago
- Knowledge cutoff
- Dec 2024
- T2 · 2 h ago
Specification
- Release date
- 25 Jul 2025
Source:OpenRouter public model & pricing listingT2observed 32 min agomedium
- Status
- deprecated
Source:Artificial AnalysisT2observed 27 min agomedium
- Openness
- open-weights
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- License
- mit
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- Architecture
- Glm4MoeForCausalLM
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- Parameters
- 110.5B
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Context window
- 128K tokens
Source:Artificial AnalysisT2observed 27 min agomedium
- Max output
- 98.3K tokens
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Knowledge cutoff
- Dec 2024
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Modalities
- text
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Input modalities
- text
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Output modalities
- text
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Languages
- en, zh
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- File size
- 220.9 GB
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- Hugging Face repo
- zai-org/GLM-4.5-Air
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Pipeline tag
- text-generation
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Model card
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Downloads
- 104,658
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 16 min agomedium
- Likes
- 635
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 16 min agomedium
- Gated
- No
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- Hf inference providers
- novita, zai-org
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- Last modified
- 2025-08-11T13:25:37+00:00
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 2 h agomedium
- Library name
- transformers
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- Aa median output tokens per second
- 59.2
Source:Artificial AnalysisT2observed 27 min agomedium
- Downloads all time
- 4,631,691
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 59 min agomedium
- Model type
- glm4_moe
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- Openrouter id
- z-ai/glm-4.5-air
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Supported parameters
- frequency_penalty, include_reasoning, max_tokens, presence_penalty, reasoning, repetition_penalty, seed, stop, temperature, tool_choice, tools, top_k, top_p
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Tags
- transformers, safetensors, glm4_moe, text-generation, en, zh
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
- Weights dtype
- BF16, F32
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
Each value shows its source, tier and observation time. Conflicting claims are kept side by side and flagged — never averaged. How AI Atlas records facts →
Provenance
Attributed facts
34
Source tiers
T234
Freshest observation
16 min ago
Conflicts
None
Modalities
- Modalities
- text
- Input
- text
- Output
- text
Capabilities
Tool calling
Yes
OpenRouter public model & pricing listing · T2
Structured output
No
OpenRouter public model & pricing listing · T2
Reasoning
Yes
OpenRouter public model & pricing listing · T2
Vision
Unavailable
Audio
Unavailable
Fine-tuning available
Unavailable
- Context window
- 128K tokens
Source:Artificial AnalysisT2observed 27 min agomedium
- Max output
- 98.3K tokens
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Knowledge cutoff
- Dec 2024
Source:OpenRouter public model & pricing listingT2observed 2 h agomedium
- Languages
- en, zh
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 1 h agomedium
| Benchmark | Score | Metric | Config | Evaluated | Source |
|---|---|---|---|---|---|
| τ²-bench | 46.49% | pass^1 | aa_slug=glm-4-5-air · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Artificial Analysis Intelligence Index | 11.09 | index | aa_slug=glm-4-5-air · version=4.3 · estimated=true | — | artificialanalysis.aiT2 |
| GPQA | 73.33% | accuracy | aa_slug=glm-4-5-air · variant=GPQA Diamond · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Humanity's Last Exam | 7.04% | accuracy | aa_slug=glm-4-5-air · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| IFBench | 37.55% | accuracy | aa_slug=glm-4-5-air · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
| Terminal-Bench | 20.45% | accuracy | aa_slug=glm-4-5-air · variant=hard · evaluator=Artificial Analysis · index_version=4.3 | — | artificialanalysis.aiT2 |
Scores are reported as published, with their evaluation configuration (harness, prompting, judge). Results with different configs are not directly comparable — see methodology.
Current prices
| Provider | Input / 1M | Output / 1M | Cached in | Batch in / out | Context | Observed | Source |
|---|---|---|---|---|---|---|---|
| Z.ai APIz-ai/glm-4.5-air | $0.13 | $0.85 | $0.025 | — | 131.1K | 32 min ago | openrouter.aiT2 |
USD per 1M tokens as published by each provider (USD). Rows are append-only: every change is kept in the history below.
Available through 1
Price history
Input price · USD / 1M tokens
Output price · USD / 1M tokens
- Z.ai API
Memory need = bytes per parameter (4-bit ≈ 0.5 × 1.15 overhead, 8-bit 1.0, fp16 2.0) + a KV-cache allowance. Not a measurement.
| Hardware | Quantization | Memory | Est. need | Fits |
|---|---|---|---|---|
| NVIDIA DGX B200 | 4bit | 1,440 GB | 64.0 GB | Yes |
| Apple M3 Ultra | 4bit | — | 64.0 GB | Yes |
| Mac Studio (Apple M5 Ultra) | 4bit | — | 64.0 GB | Yes |
| AMD Instinct MI325X | 4bit | 256 GB | 64.0 GB | Yes |
| AMD Instinct MI300X | 4bit | 192 GB | 64.0 GB | Yes |
| Apple M2 Ultra | 4bit | — | 64.0 GB | Yes |
| NVIDIA B200 | 4bit | 180 GB | 64.0 GB | Yes |
| NVIDIA H200 | 4bit | 141 GB | 64.0 GB | Yes |
| NVIDIA H200 NVL | 4bit | 141 GB | 64.0 GB | Yes |
| Apple M1 Ultra | 4bit | — | 64.0 GB | Yes |
| Apple M3 Max | 4bit | — | 64.0 GB | Yes |
| Apple M4 Max | 4bit | — | 64.0 GB | Yes |
| Mac Studio (Apple M5 Max) | 4bit | — | 64.0 GB | Yes |
| MacBook Pro (Apple M5 Max) | 4bit | — | 64.0 GB | Yes |
| NVIDIA DGX Spark | 4bit | 128 GB | 64.0 GB | Yes |
| Apple M2 Max | 4bit | — | 64.0 GB | Yes |
| NVIDIA H100 NVL | 4bit | 94 GB | 64.0 GB | Yes |
| NVIDIA A100 80GB | 4bit | 80 GB | 64.0 GB | Yes |
| NVIDIA H100 SXM | 4bit | 80 GB | 64.0 GB | Yes |
| Apple M1 Max | 4bit | — | 64.0 GB | No |
| Apple M4 Pro | 4bit | — | 64.0 GB | No |
| Mac mini (Apple M5 Pro) | 4bit | — | 64.0 GB | No |
| MacBook Pro (Apple M5 Pro) | 4bit | — | 64.0 GB | No |
| Apple M3 Pro | 4bit | — | 64.0 GB | No |
| Apple M1 Pro | 4bit | — | 64.0 GB | No |
| Apple M2 Pro | 4bit | — | 64.0 GB | No |
| Apple M4 | 4bit | — | 64.0 GB | No |
| iMac (Apple M4) | 4bit | — | 64.0 GB | No |
| Mac mini (Apple M6) | 4bit | — | 64.0 GB | No |
| MacBook Air (Apple M5) | 4bit | — | 64.0 GB | No |
| MacBook Pro (Apple M5) | 4bit | — | 64.0 GB | No |
| NVIDIA GeForce RTX 5090 | 4bit | 32 GB | 64.0 GB | No |
| Apple M2 | 4bit | — | 64.0 GB | No |
| Apple M3 | 4bit | — | 64.0 GB | No |
| NVIDIA GeForce RTX 3090 | 4bit | 24 GB | 64.0 GB | No |
| NVIDIA GeForce RTX 4090 | 4bit | 24 GB | 64.0 GB | No |
| Apple M1 | 4bit | — | 64.0 GB | No |
Ancestors 0
None recorded.
This model
GLM 4.5 Air
110.5B params
Quantizations 0
None recorded.
Descendants 1
Papers 1
- arXiv:2508.06471Active35
Repositories 0
No repositories linked yet.
As of
Rewind the record: see this entity's attributes exactly as AI Atlas knew them on a given day.
Claim history · Pipeline tag
Pipeline tagpipeline_tag1
| Value | Valid from → to | Status | Source | Confidence | Extractor |
|---|---|---|---|---|---|
| text-generation | → current | current | Hugging Face Hub (public pages, model cards, papers)T2 | medium | deterministic |
Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →
GLM 4.5 Air scores 20.45% on Terminal-Bench
artificial_analysisGLM 4.5 Air scores 46.49% on τ²-bench
artificial_analysisGLM 4.5 Air scores 37.55% on IFBench
artificial_analysisGLM 4.5 Air scores 7.04% on Humanity's Last Exam
artificial_analysisGLM 4.5 Air scores 11.09 on Artificial Analysis Intelligence Index
artificial_analysisGLM 4.5 Air: context length changed from 131072 to 128000
Context window131.1K tokens→128K tokensartificial_analysisGLM 4.5 Air: release date changed from 2025-07-20 to 2025-07-25
Release date20 Jul 2025→25 Jul 2025openrouterGLM 4.5 Air: release date changed from 2025-07-25 to 2025-07-20
Release date25 Jul 2025→20 Jul 2025huggingfaceZ.ai API lists GLM 4.5 Air at $0.13 in / $0.85 out per 1M tokens
openrouter
| Source | Document | Type | Tier | Last observed | Snapshots |
|---|---|---|---|---|---|
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/cerebras/GLM-4.5-Air-REAP-82B-A12B | model_page | T2· Quality secondary | 6 min ago | 3 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/models?author=zai-org&p=0&sort=downloads | listing | T2· Quality secondary | 16 min ago | 3 |
| Artificial Analysis | artificialanalysis.ai/leaderboards/models | leaderboard | T2· Quality secondary | 27 min ago | 1 |
| OpenRouter public model & pricing listing | openrouter.ai/api/v1/models | catalogue | T2· Quality secondary | 32 min ago | 2 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/zai-org/GLM-4.5-Air/raw/main/README.md | model_card | T2· Quality secondary | 33 min ago | 1 |
| Hugging Face Hub (public pages, model cards, papers) | huggingface.co/zai-org/GLM-4.5-Air | model_page | T2· Quality secondary | 59 min ago | 2 |
Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.