Model library / inclusionAI
Ling-3.1-flash is a 572.1 billion-parameter hybrid reasoning language model with a 262,144-token context window.
Parameters in total
Context length in tokens, as published
Weights at the released precision
Licence declared on the repository
Overview
Ling-3.1-flash is a native hybrid reasoning model released by inclusionAI. The publisher describes it as suited for complex real-world tasks such as agentic workflows, search, office productivity, software engineering, finance and materials science, and notes strong performance on benchmarks covering coding, healthcare and STEM domains.
The model contains 572.1 billion parameters, of which 25 billion are active per token, and supports up to 262,144 tokens of context. It is distributed under the MIT licence and stored in fp8 weight format, requiring substantial hardware such as multiple H200 GPUs with tensor and expert parallelism for inference.
Specification
| Repository | inclusionAI/Ling-3.1-flash |
| Publisher | inclusionAI |
| Published | 13 October 2026 |
| Architecture | BailingMoeV3ForCausalLM |
| Model type | bailing_hybrid |
| Parameters | 572,120,680,768 |
| Layers | 72 |
| Hidden size | 4,096 |
| Attention heads | 64, 64 key/value heads |
| Context length | 262,144 tokens |
| Vocabulary | 157,184 tokens |
| Quantisation | fp8 |
| Licence | mit |
Memory
Parameter count multiplied by the bytes each format uses. The key-value cache comes on top.
| Precision | Weights | Notes |
|---|---|---|
| bf16 | 1065.7 GiB | Full precision as released |
| fp8 | 532.8 GiB | 8-bit, near-lossless on most models |
| int4 | 266.4 GiB | 4-bit, smallest footprint |
The cache costs 2.2 MiB per token at 16-bit, so the full 262,144-token context of one request needs about 576.0 GiB. Concurrency multiplies that number, not the weights.
That figure is an upper bound: this model uses compressed key-value attention, so the real cache is a fraction of the figure above.
Hardware
At the released precision, with 90% of the card given to the server and 2.5 GiB kept for the runtime. The context column is what is left for the cache on that setup.
| GPU | Cards needed | Free for cache | Context it holds |
|---|---|---|---|
| RTX 4090 24GB | more than 8 | — | — |
| RTX 5090 32GB | more than 8 | — | — |
| L40S 48GB | more than 8 | — | — |
| A100 80GB | 8 | 21.8 GiB | about 9,912 tokens |
| H100 80GB | 8 | 21.8 GiB | about 9,912 tokens |
| RTX PRO 6000 96GB | 8 | 137.0 GiB | about 62,341 tokens |
| H200 141GB | 8 | 461.0 GiB | about 209,797 tokens |
Serving
A vLLM launch line for the shape above. Check the model card for a runtime the publisher recommends.
vllm serve inclusionAI/Ling-3.1-flash \ --tensor-parallel-size 8 \ --max-model-len 8192 \ --gpu-memory-utilization 0.90
Questions
The weights take about 534 GiB at the released precision. No single card in the table above holds that, so it needs several GPUs or a lower precision.
The repository declares mit. Read the licence text before commercial use: the name alone does not tell you what is allowed.
The configuration allows 262,144 tokens. Whether the whole window is usable depends on the memory left for the cache, which the hardware table works out per card.
This page is a reference, and Aiqre does not serve this model today. We deploy open-weight models on dedicated EU hardware on request, so if you want an OpenAI-compatible endpoint for it, email hello@aiqre.com with the model name and roughly what volume you expect.
Also in the library
LLaDA2.2-mini is a 16.3 billion-parameter MoE diffusion language model with 128 k token context, supporting Levenshtein editing.
Ling-3.0-flash-Fin is a 127.5 billion-parameter finance-enhanced language model for long-context financial research and analysis.
Ling-3.0-tiny is a 7.9 billion-parameter hybrid MoE language model with 1.3 billion activated parameters per token.
Ling-3.0-flash is a 127.5 B parameter hybrid-linear language model from inclusionAI for reasoning and long-context applications.
Hosting
Tell us the model and roughly what volume you expect. We reply with a price and an OpenAI-compatible endpoint, with your prompts never stored.