InferX Beta Serverless GPU Inference Platform, Built for Agent-Native Workloads

Kimi-Linear-48B-A3B-Instruct-AWQ-8bit

Kimi-Linear-48B-A3B-Instruct-AWQ-8bit is a high-efficiency, long-context model released by Moonshot AI in late 2025. It represents a significant departure from standard Transformer architectures, specifically designed to eliminate the "quadratic bottleneck" that usually slows down long-context processing.
cyankiwi text text2text
Log in to deploy: this public page shows the catalog model details, but deployment and customization stay behind login.
Log in to deploy

Metadata

Provider
cyankiwi
Modality
text
API type
text2text
Source
Created
2026-04-08 01:24:46 UTC
Updated
2026-04-08 01:24:46 UTC
Catalog version
1
Visibility
Published

Specifications

Parameters
48.00B
MoE
No
Max model length
2000
Image
vllm/vllm-openai:v0.17.0

Default Deploy Config

GPU count
2
vRAM
70000 MB
Summary
2xGPU 70000 MB

Recommended Use Cases

Model Spec