---
type: "post"
id: "inclusionai-released-ling-3-1-flash-a-560b-af9ab3"
slug: "inclusionai-released-ling-3-1-flash-a-560b-af9ab3"
title: "InclusionAI released Ling-3.1-flash , a 560B total / 25B active MoE under MIT license."
description: "**Thesis:** A large open model's practical entry point depends not only on its weights but also on whether teams can test it through familiar serving engines."
retrieval_nugget: "**Thesis:** A large open model's practical entry point depends not only on its weights but also on whether teams can test it through familiar serving engines."
published_at: "2026-10-16"
updated_at: "2026-10-16"
record_date: "2026-10-16"
date_kind: "published_at"
topics: ["open-models","models","inference"]
source_urls: ["https://huggingface.co/inclusionAI/Ling-3.1-flash"]
schema_version: "newruntime-agent-readable-v0.2"
stable_id: "post:inclusionai-released-ling-3-1-flash-a-560b-af9ab3"
status: "published"
visuals: []
editorial_provenance: {"schema_version":"newruntime-editorial-copy-v1","content_status":"source_grounded_final","final_copy_sha256":"sha256:2f003e54fe645c7ae90b5b36d5d8cac514bbecbaca2636376fccd73ad5a7ecb0","reviewed_at":"2026-10-16T20:11:23.635Z","source_evidence_count":1,"verified_claim_count":3,"site_analysis_schema_version":"newruntime-site-analysis-v1","site_object_kind":"field_note","observed_fact_count":3,"implication_count":2,"watch_condition_count":1,"related_record_count":0}
analysis: {"schema_version":"newruntime-site-analysis-v1","object_kind":"field_note","thesis":"A large open model's practical entry point depends not only on its weights but also on whether teams can test it through familiar serving engines.","observed_facts":[{"text":"The InclusionAI model page identifies Ling-3.1-flash as a 560-billion-total, 25-billion-active-parameter mixture-of-experts model.","source_urls":["https://huggingface.co/inclusionAI/Ling-3.1-flash"]},{"text":"The page lists the license as MIT.","source_urls":["https://huggingface.co/inclusionAI/Ling-3.1-flash"]},{"text":"The supplied release information also states that SGLang and vLLM have day-one support, including MTP speculative decoding.","source_urls":["https://huggingface.co/inclusionAI/Ling-3.1-flash"]}],"mechanism":"The combination of an open license and support in two established inference projects can make initial evaluation easier for teams already using those tools. The smaller active-parameter count relative to total parameters describes the model's sparse architecture, but it does not by itself establish low serving cost or high throughput.","why_now":"The release makes a new model available for evaluation, and the stated day-one integrations give builders a concrete route to test it without waiting for future engine updates.","implications":["For teams already operating SGLang or vLLM, the reported integrations may shorten the path from model discovery to a controlled trial.","The license may permit broader experimentation, subject to users verifying the license terms and their own deployment constraints."],"evidence_boundary":"These points come from the supplied public Hugging Face model page and release information. They do not independently verify model quality, benchmark rankings, latency, cost, or production readiness.","watch_conditions":["Check whether the SGLang and vLLM integrations are documented and reproducible in real deployments, and compare measured quality, throughput, and cost on representative workloads."],"related_records":[],"new_branch_reason":"No prior New Runtime record was supplied; this release therefore starts a distinct analytical record about a newly published open model and its stated serving-engine support."}
editor: {"role":"editor","name":"Andrey Reshetnikov","profile":"https://newruntime.com/owner-profile.md","profile_json":"https://newruntime.com/owner-profile.json","same_as":["https://www.linkedin.com/in/reshetnikov1/","https://t.me/lapetuse"]}
routes: {"html":"https://newruntime.com/posts/inclusionai-released-ling-3-1-flash-a-560b-af9ab3/","markdown":"https://newruntime.com/posts/inclusionai-released-ling-3-1-flash-a-560b-af9ab3.md","json":"https://newruntime.com/posts/inclusionai-released-ling-3-1-flash-a-560b-af9ab3.json"}
---

# InclusionAI released Ling-3.1-flash , a 560B total / 25B active MoE under MIT license.

## Retrieval answer

**Thesis:** A large open model's practical entry point depends not only on its weights but also on whether teams can test it through familiar serving engines.

**Thesis:** A large open model's practical entry point depends not only on its weights but also on whether teams can test it through familiar serving engines.

**Observed facts:** The InclusionAI model page identifies Ling-3.1-flash as a 560-billion-total, 25-billion-active-parameter mixture-of-experts model. The page lists the license as MIT. The supplied release information also states that SGLang and vLLM have day-one support, including MTP speculative decoding.

**Mechanism:** The combination of an open license and support in two established inference projects can make initial evaluation easier for teams already using those tools. The smaller active-parameter count relative to total parameters describes the model's sparse architecture, but it does not by itself establish low serving cost or high throughput.

**Why now:** The release makes a new model available for evaluation, and the stated day-one integrations give builders a concrete route to test it without waiting for future engine updates.

**Implications:** For teams already operating SGLang or vLLM, the reported integrations may shorten the path from model discovery to a controlled trial. The license may permit broader experimentation, subject to users verifying the license terms and their own deployment constraints.

**Evidence boundary:** These points come from the supplied public Hugging Face model page and release information. They do not independently verify model quality, benchmark rankings, latency, cost, or production readiness.

**Watch:** Check whether the SGLang and vLLM integrations are documented and reproducible in real deployments, and compare measured quality, throughput, and cost on representative workloads.

Editor: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Open Models](https://newruntime.com/topics/open-models.md): Explore the open-models topic hub.
- topic hub: [Models](https://newruntime.com/topics/models.md): Explore the models topic hub.
- topic hub: [Inference](https://newruntime.com/topics/inference.md): Explore the inference topic hub.
- collection: [Field Notes](https://newruntime.com/posts.json): Open the latest editorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)