The request names GLM-4.5-Air-FP8, a specific model. None of the candidate sources states that exact model; the closest matches are Z.ai, the maker of the GLM family, which lists GLM models on its own page 8, and AINative Studio, which offers GLM models through one OpenAI-compatible API 7. Other sources offer GLM-4V or general inference, so they are weaker fits.
A question back: Do you need GLM-4.5-Air-FP8 through a hosted API, or for self-hosting on your own GPUs?
- 1
- 2
AINative Studioainative.studiowebsite read, tools not yet checked
Offers GLM models, among others, on one OpenAI-compatible API, though no actions have been read for it yet 7.
- 3
Cheapest AI Inferencecheapestinference.comverified and signed
Provides OpenAI-compatible inference endpoints for open-source models on a flat monthly plan, with an MCP server 918.
official:MCP server · com.cheapestinference/mcp
for agents: AI Inference (OpenAI-compatible)
Sources
- 7AINative Studio | Frontier Open Models (Kimi K3, GLM, DeepSeek) — Best Price, One APIainative.studio · read 2026-10-29
- 8Z.ai - Advanced AI Chatbot & Agent powered by GLM-5.3-Flashchat.z.ai · read 2026-10-29
- 9Cheapest AI Inference — Unlimited LLM API, flat $22/mocheapestinference.com · read 2026-10-29
- 17Overview - Z.AI DEVELOPER DOCUMENT: declares "zhipu" (the record we read and signed)unfragile.ai · read 2026-10-24
- 18CheapestInference: declares "AI Inference (OpenAI-compatible)" (the record we read and signed)unfragile.ai · read 2026-10-24
For agents: the 10 actions behind this answer
cuuupid/glm-4v-9b · cuuupid/glm-4v-9b verified and signed
GLM-4V is a multimodal model released by Tsinghua University that is competitive with GPT-4o and establishes a new SOTA on several benchmarks, including OCR.
radar-get-ai-inference-summary-by-model · Cloudflare API verified and signed
Get Workers AI models summary
CreateCustomModelDeployment · Amazon Bedrock verified and signed
Deploys a custom model for on-demand inference in Amazon Bedrock.
asiryan/anything-v4.5 · asiryan/anything-v4.5 verified and signed
Anything V4.5 Model (Text2Img, Img2Img and Inpainting)
bytedance/sdxl-lightning-4step · bytedance/sdxl-lightning-4step verified and signed
SDXL-Lightning by ByteDance: a fast text-to-image model that makes high-quality images in 4 steps
huggingface-inference-api · @huggingface/inference verified and signed
run-models · run-models verified and signed
Run AI models on Replicate via predictions, webhooks, and streaming.
videoInference · Runware MCP Server found in code, not yet checked
lucataco/glm-ocr · lucataco/glm-ocr verified and signed
Compact 0.9B multimodal OCR model from Z.ai. State-of-the-art on OmniDocBench V1.5 (94.62, #1 overall). Four modes: text recognition, formula (LaTeX), table parsing, and JSON-schema information extraction. Fits on a single T4.
meta/meta-llama-3-8b-instruct · meta/meta-llama-3-8b-instruct verified and signed
An 8 billion parameter language model from Meta, fine tuned for chat completions
Labels are the instruments' verdict and are never changed by ranking. Agents and robots get this same answer as JSON at /api/v1/find?q=…&answer=1 and through the MCP server.