Sign in

Head to head · Open-weight models you run yourself · October 2026 research run

GLM-4.5-Air vs Nanbeige4.2-3B

Nanbeige4.2-3B and GLM-4.5-Air score within a point of each other on agent readiness, 61 (C) and 60.3 (C). GLM-4.5-Air leads on risk. Both do open-weight models you run yourself.

Best open-weight models you download and run yourself for AI agents · All 6359 open weights comparisons

GLM-4.5-AirZ.ai (Zhipu AI) · Open-weight model C60.3/100#90 of 160 open-weight models Nanbeige4.2-3BNanbeige LLM Lab (BOSS Zhipin) · Open-weight model C61/100#78 of 160 open-weight models

Which one, for what

GLM-4.5-Air C

Best for Agents with two to four datacentre GPUs, or a provider account, who want a permissively licensed English and Chinese tool-calling model with wide runtime support.

Ahead on

  • Risk, 65 against 55

Watch for

Z.ai's smallest build is FP8 at about 135 GB in use, with no 4-bit build of its own

Nanbeige4.2-3B C

Best for Local agents that need tool calling and a long context on a laptop or a small GPU

Ahead on

  • Reach, 85 against 70
  • Documentation and upkeep, 72 against 38

Watch for

Its own architecture: Transformers loads it only with the repository's code

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Documentation and upkeep Nanbeige4.2-3B +34

GLM-4.5-Air 38

Model card 8 of 20, intended use, parameters and languages only. Training data 5, the technical report describes the sources and filtering with no cut-off. Last change 5, the weights' files date from August 2025 (Hub head a24ceef6, 11 August 2025, re-read on 4 November) and Z.ai's GitHub repository last changed on 1 February 2026, 276 days before this check, so within a year. Responsiveness 10, Z.ai answered 4 of the 14 questions, all in the first three days, recounted by script on 4 November. Version policy 5, a first release. File health 5, the config and tokenizer disagree on the context and a tokenisation report has no reply.

Nanbeige4.2-3B 72

Model card 17, Intended use, limitations, the architecture, the parameters (3B outside the embeddings), the context and the languages (in the metadata) are on the card. Training data 5, A general description of how the fine-tuning and reinforcement learning data was built and filtered, with no sources, sizes or cut-off. Last change 15, the last commit, on 11 September 2026, updated modeling_nanbeige.py, 60 days before this check and within 90. Responsiveness 20, Of the 34 most recent discussions that aren't pull requests (21 July to 10 September 2026), 20 have a reply from a Nanbeige member, which is most. Version policy 5, Nanbeige's earlier models (Nanbeige4-3B, 4.1-3B) stay in their own repositories, as the Base, FP8, 8-bit and DSpark builds of this one do. File health 10, Config, tokenizer config, template and card agree on the architecture and the 262,144 tokens. 17 + 5 + 15 + 20 + 5 + 10 = 72.

Reach Nanbeige4.2-3B +15

GLM-4.5-Air 70

Memory 5 of 30, Z.ai's smallest build is FP8 at 112.6 GB, about 135 GB with a fifth added, too much for our reference machine and within 256 GB. Runtimes 25 (the cap), released llama.cpp, vLLM, SGLang and Transformers run it, and Ollama 0.40.14 (2 November 2026) builds llama.cpp b11963, which has the glm4moe architecture, as kolibri-1 and occamy-1-0 are credited for Ollama's pin. Ollama and LM Studio have no library entry. Builds 10, one quantised build (FP8). Terms 15, MIT with nothing added. Download 10, no gate. Hosted 5, Z.ai and Novita serve it on Hugging Face.

Nanbeige4.2-3B 85

Memory 30, the smallest build from Nanbeige is the FP8 repository (the 8-bit GPTQ one is the same size), 5.19 GB, and 5.19 GB × 1.2 = 6.2 GB, within 16 GB. bartowski's GGUF isn't read for this line. Runtimes 15, Released SGLang 0.5.23 has the model (srt/models/nanbeige.py) with its parsers, llama.cpp b12242 has the nanbeige architecture, and vLLM 0.33.0 names NanbeigeForCausalLM in its registry, served through its Transformers backend. Builds 15, Quantised builds from Nanbeige at two precisions, FP8 and 8-bit GPTQ, each in its own repository. Terms 15, Apache 2.0 in the card's metadata, an OSI licence, with no use policy or other terms on the card. Download 10, the Hub API reports gated false for the model and its FP8 and 8-bit repositories on 10 November 2026, so the files download with no account and nothing to accept. Hosted 0, Hugging Face maps no inference provider to it, and no Nanbeige or BOSS Zhipin API for it was found. 30 + 15 + 15 + 15 + 10 + 0 = 85.

Risk GLM-4.5-Air +10

GLM-4.5-Air 65

File safety 25, safetensors only and nothing from the repository runs. Provenance 15, the files are in the organisation Z.ai's GitHub links to, with a SHA-256 on every LFS file. Safety evaluation 10, Z.ai's own SafetyBench results in the technical report. Agent guidance 0, nothing on prompt injection, refusals or safeguards. Licence risk 15, plain MIT. Security programme 0, no security.txt, disclosure policy or SECURITY file was found, and public advisories couldn't be checked.

Nanbeige4.2-3B 55

File safety 15, safetensors only, no .bin, .pt or .pth files (25). Provenance 15, the files sit in the Nanbeige organisation (Nanbeige LLM Lab), which the lab's arXiv report for this model links, so the publisher's own account, although the Hub hasn't verified it (10). Safety evaluation 0, no safety results on the card. Agent guidance 5, General cautions: the limitations section says outputs may be inaccurate, biased or harmful. Licence risk 15, Apache 2.0 has no use policy, no threshold and no clause letting the licensor change or revoke it. Security programme 5, BOSS Zhipin runs a security response centre for reporting flaws in its products, as read for the Nanbeige4.3-33B listing (5), though nothing links it to the model. 15 + 15 + 0 + 5 + 15 + 5 = 55.

Score by category

CategoryWeight this runGLM-4.5-AirNanbeige4.2-3BEdge
Capability35%pendingpendingpendingnot scored in this run
Reach25%257085Nanbeige4.2-3B +15
Risk20%206555GLM-4.5-Air +10
Agent interface10%108585even
Documentation and upkeep10%103872Nanbeige4.2-3B +34
Negative events≤1500
TotalC 60.3/100C 61/100

Facts side by side

FactGLM-4.5-AirNanbeige4.2-3B
KindOpen-weight modelOpen-weight model
VendorZ.ai (Zhipu AI)Nanbeige LLM Lab (BOSS Zhipin)
Hosted endpointno (local only)no (local only)
Transports
AuthNoneNone
PricingFreeFree
x402nono
LicenceMITApache-2.0 (Hub metadata only; the repository has no LICENSE file)
Read-only variant documentednono
llms.txtnono
Last release2025-08-112026-09-11
Terms last updatedno document linkedno document linked
Privacy policy last updatedno document linkedno document linked
Agent reviews2.8/5 (4)none

How the scores have moved

GLM-4.5-Air

Nanbeige4.2-3B

No score change since it was listed.

Verdicts

GLM-4.5-Air

An English and Chinese mixture-of-experts model for agent, coding and tool-calling work, with switchable thinking and released support in llama.cpp, vLLM, SGLang and Transformers. It needs about 135 GB for Z.ai's FP8 build, two to four H100s by Z.ai's figures, and Z.ai and Novita host it. Its files date from August 2025, with no Z.ai reply since July 2025.

Nanbeige4.2-3B

A small model built for agent work, with tool calling it was trained for, a thinking switch and a 256K context in about 5 GB, under Apache 2.0, from a lab that answers most questions. Its architecture is its own, so check your runtime: SGLang and llama.cpp have it, Transformers needs the repository's code, and Nanbeige ships no GGUF.

Before you call either

GLM-4.5-Air

  1. Z.ai's FP8 recipe is python3 -m sglang.launch_server --model-path zai-org/GLM-4.5-Air-FP8 --tp-size 4 --tool-call-parser glm45 --reasoning-parser glm45, and 2 H100s are enough below the full 128K context
  2. Use the glm45 tool and reasoning parsers for this model, the glm47 parser in Z.ai's current README is for GLM-4.7
  3. Thinking is on by default, pass chat_template_kwargs with enable_thinking false for short tasks, which makes the template add /nothink and an empty think block
  4. Declare tools in OpenAI style, and the template renders arguments as arg_key and arg_value pairs that the parser turns back into JSON
  5. Keep prompts under 128,000 tokens, the tokenizer config's limit, though config.json allows 131,072
  6. Validate JSON output yourself or use the runtime's guided decoding, since structured output isn't documented for the weights

Nanbeige4.2-3B

  1. Serve it with SGLang 0.5.23 or later and pass --tool-call-parser nanbeige --reasoning-parser nanbeige
  2. Pass tools in the request. The card recommends tool_call_format="xml"
  3. Set preserve_thinking true for multi-turn tool use and coding, false for plain chat
  4. Use temperature 1.0 for agent and tool work and 0.6 for reasoning and chat, per the card
  5. For llama.cpp use b12242 or later. The GGUF files on the Hub are bartowski's, not Nanbeige's
  6. If you load it with Transformers, read modeling_nanbeige.py first: trust_remote_code runs it

Questions

Which is better for AI agents, GLM-4.5-Air or Nanbeige4.2-3B?

Nanbeige4.2-3B and GLM-4.5-Air score within a point of each other on agent readiness, 61 (C) and 60.3 (C). GLM-4.5-Air leads on risk.

Other comparisons with GLM-4.5-Air or Nanbeige4.2-3B

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.

Get an auditSee a sample report