AI Infra Interviews logo

Ling / Hybrid KDA / gated MLA MoE

Ling-3.1-flash

Count recurrent and latent state separately

Checkpoint uploadSources checked 2026-11-16

A cluster candidate for text agents when its task quality justifies the expert store. Its smaller active path reduces computation without making all weights disappear.

560Breported total parameters
25Bactive parameters per token
262,144configured token positions

Publisher-reported rounded count; the pinned tensor index is a separate payload measurement.

Evaluate it for

Text-agent evaluation with a runtime that supports this exact hybrid layout.

Choose another path when

A workstation budget, image inputs, or a requirement for already-qualified million-token concurrency.

Release context

Why this release matters

Sixty-three Kimi Delta Attention layers retain recurrent state; nine gated multi-head latent-attention layers retain history. Cache reuse must restore both.

Dated weight-upload commit; not inferred from repository creation or an API listing.

Release evidence · Follow the Ling timeline →

Deployment starting points

Which hardware, how many, at what precision?

Start with an exact artifact and an engine that supports it. A configuration below is evidence of a documented path; its device count does not promise a particular throughput or concurrency.

573.75 GB of tensor data in the pinned index

This is the index’s reported tensor payload, in decimal GB. It excludes file headers and runtime memory. Mixed precision, conversion and host/device placement determine how much GPU memory the loaded model needs.

Inspect the exact index metadata →
Capacity screen: would the weights fit?

This estimates inference memory for a hypothetical uniform precision. It is useful for rejecting an allocation that is too small. It does not establish a working deployment or estimate training memory.

A uniform-precision GPU count would hide this model’s component layout. Publisher-reported rounded count; the pinned tensor index is a separate payload measurement. Use the documented artifact and placement path above; no GPU count is inferred here.

Architecture in practice

What the serving engine has to do

Two kinds of layers retain two kinds of state.

A token enters a layerthe layer type selects itspathRecurrent layerupdate a compact stateAttention layerread retained tokenhistoryRecurrent statemaintained for thissequenceAttention cachelayout depends on themodel

Hybrid attention does not mean zero request state.

A full-attention formula cannot describe every layer.

Schematic of a hybrid attention stack, not a literal layer ordering. Recurrent layers update a compact state; attention layers retain their own history. Both kinds of state matter for prefix reuse and concurrent requests.

Mechanism schematic based on the pinned model card and configuration. It explains a design principle; it is not a full implementation graph.

Inspect the full checkpoint specifications
Inspected fields for this exact revision. KV heads alone do not describe latent, hybrid or shared-cache layouts.
Exact checkpointinclusionAI/Ling-3.1-flash
Text model typebailing_hybrid
Layers72
Hidden width4,096
Attention / KV heads64 / 64
Head dimensionNot verified
Routed / selected experts512 / 8
Configured positions262,144
Documented extensionNot recorded for this checkpoint
Inputs → outputtext → text
License metadataapache-2.0

63 KDA layers; 9 gated MLA layers. These counts describe the configured pattern.

Open weights do not imply unrestricted use. Read the applicable license terms linked from the pinned card.

A useful comparison

What else belongs on the shortlist?

Compare the two Ling revisions under one harness. The smaller alternative changes residency costs as well as possible task quality.

27 Oct 2026

Ling-3.1-flash-1019

A post-training update needs its own deployment evidence

560B reported total · text

23 Oct 2026

MiniMax-M3.1-Flash

M3 with sparse attention everywhere and a quantized-trained store

428B reported total · text + image + video

These are editorial comparison candidates. We have not run a matched quality or serving benchmark, so this is not a ranking.

Prepare to explain it

The interview lesson

Budget the 573.75 GB tensor payload separately from recurrent snapshots and the nine latent caches.

Attention layouts · Quantization · Courses and worked examples

For a deployment evaluation, record exact weights, precision, engine, device count, interconnect, prompt/output lengths and concurrency. Report task success, errors, TTFT, TPOT and useful throughput together.

Sources and citation

Our assessment is editorial judgment based on the linked architecture and deployment evidence. Model facts are publisher-reported or attributed to runtime maintainers; no independent GPU benchmark was run.

AI Infra Interviews, “Ling-3.1-flash: release, architecture and deployment”, checked 2026-11-16. Preserve this date and the exact checkpoint when citing.

Model and research notes as JSON · InclusionAI profile