paper7Updated Nov 20, 2026 · Português
← All organizations

Mangaba AI

location not provided
Small companyruns on device

Portuguese text LLMs: models trained from scratch (Nano-1M, Base-91M), 8B vertical specialists and GGUF; sells training, fine-tuning, agents, Router and Brier.

Models
35
Datasets
2
Downloads in 30 days
611
Followers
3
Likes
1
Smallest model
1M

Releases per quarter

Models and datasets, by creation date on Hugging Face.

2023202420252026

Model licenses

What each model card declares

  • apache-2.025
  • other9
  • gemma1

Dataset licenses

Commercial use under the declared license

  • Allowed1
  • No license1

Model types

By declared task

  • no task21
  • text-generation12
  • tabular-classification2

Where the models run

By file format and size

  • server23
  • device12

Languages

Declared in model and dataset cards

  • pt33
  • en16

Models on Hugging Face (35)

Only what is published under the organization's Hugging Face account. Products and models under other accounts are not listed here.

ModelTaskRuns onSizeLicenseLikesUpdatedDownloads
Mangaba-Agro-8Btext-generationserverapache-2.00today0
Mangaba-Financeiro-8Btext-generationserverapache-2.00today0
Mangaba-Juridico-8Btext-generationserverapache-2.00today0
Mangaba-Licitacoes-8Btext-generationserverapache-2.00today0
Mangaba-Saude-Suplementar-8Btext-generationserverapache-2.00today0
Mangaba-Telecom-8Btext-generationserverapache-2.00today0
Mangaba-Trabalhista-8Btext-generationserverapache-2.00today0
Mangaba-Tributario-8Btext-generationserverapache-2.00today0
Mangaba-S5-nano-LoRAtext-generationdeviceapache-2.009 days ago0
Mangaba-Agente-Ferramentas-30Bserverapache-2.0014 days ago0
Mangaba-Bolt-Ferramentas-4Bserverother014 days ago0
Mangaba-Flash-Ferramentas-1.2Bserverother014 days ago0
Mangaba-Leve-Ferramentas-1.7Bserverapache-2.0014 days ago0
Mangaba-Zero-Ferramentas-14Bserverapache-2.0014 days ago0
Mangaba-Zero-Ferramentas-32Bserverapache-2.0014 days ago0
Mangaba-AI-Nordeste-4Btext-generationdeviceapache-2.0015 days ago0
Mangaba-Agente-Rapido-Ferramentas-8Bserverother015 days ago0
Mangaba-Base-91Mtext-generationdevice91Mapache-2.0015 days ago146
Mangaba-Nano-1Mtext-generationdevice1Mapache-2.0015 days ago123
Mangaba-mini-Ferramentas-4Bserverother015 days ago0
jepa-fraude-ulbtabular-classificationserverother015 days ago0
jepa-sigreg-ulbserverother018 days ago0
jepa-fraude-ulb-weightstabular-classificationserverother023 days ago0
brier-v5serverother01 month ago0
brier-zeroserverother01 month ago0
brier-v2serverapache-2.001 month ago9
brier-v3serverapache-2.001 month ago23
mangaba-reforma-tributaria-2bdevicegemma13 months ago0
Mangaba-GPT-OSS-20B-GGUFdeviceGGUFapache-2.004 months ago82
Mangaba-Gemma-4-12B-it-GGUFdeviceGGUFapache-2.004 months ago21
Mangaba-Gemma-4-26B-A4B-it-GGUFdeviceGGUFapache-2.004 months ago78
Mangaba-Qwen3-Coder-30B-A3B-GGUFdeviceGGUFapache-2.004 months ago18
Mangaba-Qwen3.5-9B-GGUFdeviceGGUFapache-2.004 months ago16
Mangaba-Qwen3.6-27B-GGUFdeviceGGUFapache-2.004 months ago11
Mangaba-Qwen3.6-35B-A3B-GGUFdeviceGGUFapache-2.004 months ago31

Datasets (2)

Commercial use reflects only the declared license; an open license does not guarantee that the content owners granted the rights.

DatasetLicenseCommercial useRowsUpdatedDownloads
mangaba-pro-122b-benchmarkapache-2.0Allowedn<1K2 months ago41
reforma-tributaria-instructno licenseNo license1K<n<10K3 months ago12

Declared links

Only what the authors declared in each model card. A missing declaration does not mean there is no use.

Trains on data from
Polygl0t (gigaverbo-v2, gigaverbo-v2-synth); David-Egea (Creditcard-fraud-detection)
Builds on models from
Qwen (Qwen3-8B, Qwen3-30B-A3B-Instruct-2507, Qwen3-1.7B and 8 more); mlx-community (Qwen3-4B-Instruct-2507-4bit, gemma-2-2b-it-4bit); google (gemma-4-12B-it, gemma-4-26B-A4B-it); Maritaca AI (S5-nano); nvidia (NVIDIA-Nemotron-3.5-Bolt-E4B-BF16); LiquidAI (LFM2.5-1.2B-Instruct); inclusionAI (Ling-3.0-tiny); openai (gpt-oss-20b)