Skip to main content

Mind LabLanguage modelWeights available

Macaron-V1-Venti

PopularityRank 292 of 339 language models

21out of 100first value

usage data only, range 7 to 35

Summary

Macaron-V1-Venti combines a GLM-5.2 base with four LoRA specialists for chat, agent tasks, coding, and Generative UI. A router selects a specialist for each new request.

The open-weights checkpoint is available for local use. Its official API ID is macaron-v1-venti. Mind Lab released V1 on July 21, 2026.

Popularity compared

How often the model is searched for, used and mentioned, compared with every model of its category.

Popularity index

The 100 most popular models of this category, 0 to 100, higher is more popular

20 of 101 models

As of November 2026 · Source: gradually.ai popularity index

Over time

First value from November 16, 2026. The timeline appears from the second week.

What is gaining popularity right now, across models, vendors, harnesses and AI subscriptions, is collected in All AI trends.

All specifications

Architecture

Checkpoint
mindlab-research/Macaron-V1-Venti
Source: Checkpoint
Base model
GLM-5.2
Source: Base model
Architecture
Routed Mixture-of-LoRA architecture with specialists for chat, agent tasks, coding, and Generative UI.
Source: Architecture

Input and output

Input
Text
Source: Input
Output
Text
Source: Output
Context window
1,048,576 positions in the checkpoint configuration; the model card states 1 million tokens.
Source: Context window

Watermarking and provenance

AI watermarking refers to invisible signals embedded in the output. Visible marks and provenance metadata are listed separately. Unknown means there is insufficient evidence for that particular field.

AI watermarking
Unknown
no source
Watermarking technique
Not known
no source

Licence and access

Weights license
MIT license according to the model card. Any applicable base-model and dependency requirements must also be observed.
Source: Weights license
Shutdown
None announced
no source

API

Mint API model ID
macaron-v1-venti
Source: Mint API model ID

Further details

Published model scale
Mind Lab gives 748B as the release label: a 744B GLM-5.2 base plus four LoRA specialists labeled 1B each. The model card explains that the stored adapter tensors sum to more and that the release figure is not an unambiguous total parameter count.
Source: Published model scale

Frequently asked questions about Macaron-V1-Venti

How large is the context window of Macaron-V1-Venti?

1.05 M tokens. That is how much text Macaron-V1-Venti reads at once, about 1,570 A4 pages.

Page count as a rule of thumb from Artificial Analysis

Context windowTokens the model reads at once · higher is better
Macaron-V1-VentiRank 12 of 1121.05M
Gemini 4 Carbon1.05M
Gemini 4 Argon1.05M
Muse Wonder1.05M
Kimi K3.1 Pro1.05M
Muse Ember 1.01.05M
Gemini 3.7 Flash1.05M

Can I run Macaron-V1-Venti locally?

Yes, the weights are open.

Stated by Mind Lab

Open weights
yes
License
MIT license according to the model card. Any applicable base-model and dependency requirements must also be observed.
Parameters
n/a
active per token
n/a

Sources and data date

Show all 11 sources
TypeEvidence and data date
Input

Documented modalities. Retrieved November 16, 2026.

Output

Documented modalities. Retrieved November 16, 2026.

Additional sourceMind Lab Macaron-V1 release announcement (retrieved November 16, 2026) · Mind Lab Macaron-V1 release announcement · Editorial description reviewed November 16, 2026
Additional sourceMacaron-V1-Venti checkpoint configuration (retrieved November 16, 2026)
Additional sourceMint Recursive hosted model IDs (retrieved November 16, 2026) · Mint Recursive hosted model IDs · Editorial description reviewed November 16, 2026
Additional sourceMind Lab Macaron-V1 Hugging Face collection (retrieved November 16, 2026)
Additional sourceOpenRouter: Popularity index (derived), CC BY 4.0
Additional sourceVercel AI Gateway: Popularity index (derived), CC BY 4.0
Additional sourceOllama: cumulative downloads for the popularity index
Additional sourceReplicate: cumulative runs for the popularity index
Additional sourceGoogle Trending Now: limited feed selection for Germany and the US
Additional sourceCreative Commons Attribution 4.0