> ## Documentation Index > Fetch the complete documentation index at: https://developer.eka.care/llms.txt > Use this file to discover all available pages before exploring further. # Models # Open Models for Indian Healthcare AI Eka Care releases purpose-built models for speech, documents and text in Indian clinical settings. All models are in the [EkaCare Public Healthcare LLMs](https://huggingface.co/collections/ekacare/ekacare-public-healthcare-llms) collection on Hugging Face and on the [India AI AIKosh platform](https://aikosh.indiaai.gov.in). Try them live at [medai.eka.care](https://medai.eka.care/). | Model | Task | Size | Links | | - | - | - | - | | Parrotlet-a 2.5 Pro | Medical ASR (5 Indian languages) | \~5B | [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_a_2p5_pro.html) · [Hugging Face](https://huggingface.co/ekacare/parrotlet-a-2.5-pro) | | Parrotlet-A-EN-5B | Medical ASR (English) | 5B | [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_a_en_5b.html) · [Hugging Face](https://huggingface.co/ekacare/parrotlet-a-en-5b) | | Parrotlet-v 2.5 Pro | Medical record extraction (vision) | \~2B | [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_v_2p5_pro.html) · [Hugging Face](https://huggingface.co/ekacare/parrotlet-v-2.5-pro) | | Parrotlet-V-Lite-4B | Medical record parsing (vision) | 4B | [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_v_lite_4b.html) · [Hugging Face](https://huggingface.co/ekacare/parrotlet-v-lite-4b) | | Med Doc Classifier | Medical document classification | \~96M | [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/med_doc_classifier.html) · [Hugging Face](https://huggingface.co/ekacare/med-doc-classifier) | | Document PII Redactor | PII detection & redaction | – | [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/document_pii_redactor.html) · [Hugging Face](https://huggingface.co/ekacare/document-pii-redactor) | | Parrotlet-E | Multilingual medical embeddings | – | [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_e.html) · [Hugging Face](https://huggingface.co/ekacare/parrotlet-e) | *** ## Speech ### Parrotlet-a 2.5 Pro * A purpose-built ASR model for medical speech in Indian healthcare settings, transcribing Indian English, Hindi, Marathi, Kannada and Telugu, including the heavily code-mixed speech typical of real consultations. * Combines a Whisper large-v3 encoder with a MedGemma 4B decoder, further tuned with GRPO on medical conversation data. Best Semantic WER in every language against nine production ASR systems on medical conversation sets. * [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_a_2p5_pro.html) | [Hugging Face](https://huggingface.co/ekacare/parrotlet-a-2.5-pro) | [Live demo](https://medai.eka.care/asr) | [Launch blog](https://info.eka.care/services/benchmark-update-releasing-parrotlet-a-2-5-pro) *** ### Parrotlet-A-EN-5B * A healthcare-optimized ASR model for transcribing English medical speech in Indian clinical settings, combining a Whisper V3 encoder with a MedGemma3 4B decoder. * Designed for accurate short-form medical audio transcription, it enables high-fidelity speech-to-text conversion for scribe and voice-enabled healthcare applications. * [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_a_en_5b.html) | [Hugging Face](https://huggingface.co/ekacare/parrotlet-a-en-5b) | [Launch blog](https://info.eka.care/services/parrotlet-a-en-5b-releasing-our-purpose-built-llm-for-english-asr-in-indian-healthcare) *** ## Documents & Vision ### Parrotlet-v 2.5 Pro * A fine-tune of Gemma 4 E2B (\~2B effective parameters) for structured, schema-driven extraction from Indian medical records: prescriptions, invoices, lab reports and discharge summaries, plus PII detection. * Prompted with a document image and a single task tag; returns schema-conformant JSON. Released under Apache 2.0 in bf16. * [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_v_2p5_pro.html) | [Hugging Face](https://huggingface.co/ekacare/parrotlet-v-2.5-pro) | [Live demo](https://medai.eka.care/medical-records) | [Launch blog](https://info.eka.care/services/releasing-parrotlet-open-models-open-weights-for-medical-document-intelligence) *** ### Parrotlet-V-Lite-4B * A purpose-built vision LLM optimized for parsing Indian medical records, including lab reports, digital prescriptions, document classification, and PII extraction. * Deployable via Hugging Face, it enables structured data extraction from medical images and PDFs for healthcare AI and interoperability workflows. * [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_v_lite_4b.html) | [Hugging Face](https://huggingface.co/ekacare/parrotlet-v-lite-4b) | [Launch blog](https://info.eka.care/services/parrotlet-v-lite-4b-releasing-our-purpose-built-vision-llm-for-parsing-medical-records) *** ### Med Doc Classifier * A lightweight (\~96M parameter) encoder-only model on a SigLIP2-base backbone that sorts a document image into one of 27 classes across two hierarchical levels, and also flags whether it is medical and whether it is handwritten. * Built by training many specialised models, ensembling them, and distilling the ensemble into one compact student. Released under Apache 2.0. * [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/med_doc_classifier.html) | [Hugging Face](https://huggingface.co/ekacare/med-doc-classifier) | [Live demo](https://medai.eka.care/classifier) | [Launch blog](https://info.eka.care/services/releasing-parrotlet-open-models-open-weights-for-medical-document-intelligence) *** ### Document PII Redactor * Model weights powering the open-source [document-pii-redactor](https://github.com/eka-care/document-pii-redactor), which detects and redacts PII in both text and document images and is light enough to run on CPU. * Trained to understand Indian names, documents and contexts; the text model works across Indian languages. The core contribution is the PII token classifier, with OCR as a pluggable front-end. * [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/document_pii_redactor.html) | [Hugging Face](https://huggingface.co/ekacare/document-pii-redactor) | [GitHub](https://github.com/eka-care/document-pii-redactor) | [Live demo](https://medai.eka.care/pii-redactor) | [Launch blog](https://info.eka.care/services/releasing-parrotlet-open-models-open-weights-for-medical-document-intelligence) *** ## Embeddings ### Parrotlet-E * A state-of-the-art multilingual embedding model optimized for entity-level medical concept representation across 12 Indic languages and English. * Fine-tuned on 18M+ SNOMED CT and UMLS-aligned term pairs, it enables robust cross-lingual medical coding, semantic search, and retrieval with benchmark-leading performance on Eka-IndicMTEB. * [AIKosh](https://aikosh.indiaai.gov.in/home/models/details/parrotlet_e.html) | [Hugging Face](https://huggingface.co/ekacare/parrotlet-e) | [Launch blog](https://info.eka.care/services/introducing-parrotlet-e-and-eka-indicmteb-bridging-indias-multilingual-healthcare-gap) *** ## Evaluation Models are evaluated with [KARMA – OpenMedEvalKit](https://github.com/eka-care/karMA-OpenMedEvalKit), our open-source evaluation framework, on the [Eka evaluation datasets](/eka-medai/datasets). Current results are on each model card and in the launch blogs linked above; the [Eka Medical ASR dataset card](https://huggingface.co/datasets/ekacare/eka-medical-asr-evaluation-dataset) carries a public leaderboard. *** *Models are released under the license stated on their respective Hugging Face model cards. Some models require accepting the terms on Hugging Face before download.* This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.