Local LLM model page

Ling 1T (1T MoE)

Ant Group / InclusionAI trillion-param MoE. 50B active per token, 128K context. Strong Chinese + English, open weights with commercial licence. Tops many bilingual benchmarks. Datacenter-only.

Find the best model for my hardwareBrowse all 183 LLMs
Parameters
1T (50B active)
Minimum RAM
1024 GB
Model size
620 GB
Quantization
Q4_K_M

Can Ling 1T (1T MoE) run locally?

Ling 1T (1T MoE) is best suited for server-grade or multi-GPU systems. LocalClaw recommends Q4_K_M as the default quantization, with at least 1024 GB RAM.

Search term for LM Studio or compatible runtimes: ling-1t

Hugging Face repository: inclusionAI/Ling-1T

chatcodereasoningquality

Strengths

  • Ant Group / InclusionAI trillion-param MoE. 50B active per token, 128K context. Strong Chinese + English, open weights with commercial licence. Tops many bilingual benchmarks. Datacenter-only.

Limitations

  • Performance depends heavily on quantization, RAM bandwidth and runtime support.

Best use cases

  • chat
  • code
  • reasoning
  • quality

Benchmarks

Speed: 2/10

Quality: 10/10

Coding: 9/10

Reasoning: 9/10

Technical details

Developer: ling

License: See model repository

Context window: Unknown tokens

Architecture: See model card

Released: 2025-10