What are the hardware requirements for Ling 2.6 Flash?

Ling 2.6 Flash requires approximately 208.2 GB of VRAM and supports a context window of 128K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Ling 2.6 Flash and how large is it?

Ling 2.6 Flash is developed by inclusionAI, a model with 107.5b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Ling 2.6 Flash perform on standard benchmarks?

Ling 2.6 Flash has the following published scores: LMSYS ELO 1484. Compare against reference models on this page or on the LLM Explorer leaderboards.

Ling 2.6 Flash by inclusionAI — VRAM 208.2GB, 128K context

Name: Ling 2.6 Flash
Author: inclusionAI

Ling 2.6 Flash is an open-source language model by inclusionAI. Features: 107.5b LLM, VRAM: 208.2GB, Context: 128K, License: mit, LLM Explorer Score: 0.49, ELO: 1484.

Arxiv:2606.15079 Bailing hybrid Conversational Custom code En Eval-results Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/inclusionAI/Ling-2.6-flash

Ling 2.6 Flash Benchmarks

LMSys ELO 1484 vs 1272 (so35) 16.7%

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

Ling 2.6 Flash Parameters and Internals

LLM Name	Ling 2.6 Flash
Repository 🤗	https://huggingface.co/inclusionAI/Ling-2.6-flash
Model Size	107.5b
Required VRAM	208.2 GB
Updated	2026-07-20
Maintainer	inclusionAI
Model Type	bailing_hybrid
Model Files	7.8 GB: 1-of-26 10.7 GB: 2-of-26 8.6 GB: 3-of-26 6.5 GB: 4-of-26 6.8 GB: 5-of-26 7.7 GB: 6-of-26 6.4 GB: 7-of-26 6.5 GB: 8-of-26 6.5 GB: 9-of-26 10.7 GB: 10-of-26 8.6 GB: 11-of-26 10.7 GB: 12-of-26 10.0 GB: 13-of-26 10.7 GB: 14-of-26 8.6 GB: 15-of-26 6.5 GB: 16-of-26 6.5 GB: 17-of-26 10.7 GB: 18-of-26 9.9 GB: 19-of-26 6.4 GB: 20-of-26 6.4 GB: 21-of-26 6.5 GB: 22-of-26 6.5 GB: 23-of-26 6.5 GB: 24-of-26 7.7 GB: 25-of-26 7.8 GB: 26-of-26 6.6 GB
Supported Languages	en
Model Architecture	BailingMoeV2_5ForCausalLM
License	mit
Context Length	131072
Model Max Length	131072
Transformers Version	4.56.2
Tokenizer Class	PreTrainedTokenizerFast
Padding Token	<\|endoftext\|>
Vocabulary Size	157184
Torch Data Type	bfloat16

Rank the Ling 2.6 Flash capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Quantized Models of the Ling 2.6 Flash

Model	Likes	Downloads	VRAM
Ling 2.6 Flash Mlx 8bit	1	69	111 GB
Ling 2.6 Flash Mlx 5bit	1	45	71 GB
Ling 2.6 Flash Mlx 6bit	1	36	85 GB
Ling 2.6 Flash MLX Q9	3	182	117 GB
Ling 2.6 Flash Mlx 4bit DWQ	7	82	58 GB
Ling 2.6 Flash MLX 9bit	2	91	117 GB
Ling 2.6 Flash Mlx 4bit	2	68	58 GB
Ling 2.6 Flash Mlx 4bit Gs32	2	22	65 GB

Best Alternatives to Ling 2.6 Flash

Best Alternatives	Context / RAM	Downloads	Likes
Ling 2.6 Flash Base	256K / 109.2 GB	330	14