Model reference · open weights

meiki.txt.recognition

meiki.txt.recognition is an open-weight language model from rtr46, listed in the AxForge catalogue. AxForge can bring it up on EU-owned hardware for you on request — with the licence handled where one is required.

LLMs rtr46 1 variants 101k downloads/mo
Request this model on EU hardware All served models Not on the shared API today — deployed on request.

About

What meiki.txt.recognition is

Official GitHub Repository: meikiocr This model is a core component of the meikiocr pipeline. For the full implementation, command-line script, and documentation, please see the official GitHub repository. meiki.text.recognition.v0 pareto-optimal text recognition model. trained on japanese video games. meiki.text.recognition achieves state-of-the-art text recognition accuracy as well as latency by redefining "text recognition" as "character detection". the model is a fine-tune of https://github.com/Peterande/D-FINE object detecor combined with a mobilenetv4 CNN backbone. to my knowledge there is no existing, open weight text recognition model with a better accuracy/latency tradeoff for japanese text recognition. intended use and constraints - it is specifically trained on japanese video games, therefore performance may vary outside of this use case - input needs to be resized and padded to be 960x32px - outputs detection results as characters + bbox + confidence. check the inference.py script for a suggested post processing algorithms - can detect up to 48 characters - only works on horizontal text benchmarks how to use please refer to this demo inference script: https://huggingface.co/rtr46/meiki.txt.recognition.v0/blob/main/inference.py examples releases - 21.02.2026: i released a new version of the checkpoint that should improve accuracy. if you want to use the old one for some reason use this revision: ddd06176a4da56fba082293dbe9898d4e5998af2

Summarised from the published model card. Read the full card on the HuggingFace links below.

Specifications

What it is

Makerrtr46
TypeLanguage models
Variants1
Runs withtensorrt
Released2025-11-03
Popularity101k downloads / month
Likes7
LicenceOpen, with conditions

How it works

How language models work

Your prompttext / messagesTransformerattention over tokensNext-token loopgenerate + streamResponsetext · tool callsA language model reads your tokens and predicts the next one, again and again, streaming the reply back.

Variants

Sizes & precisions

Open weights ship in several sizes and precisions. One page, all the variants — pick the one that fits your GPU. VRAM figures are estimates from model size.

VariantParamsPrecisionVRAMFits 16 GBWeights
meiki.txt.recognition.v0BF16Weights ↗

Using it via the API

Call it like any OpenAI endpoint

Once AxForge deploys meiki-txt-recognition for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (meiki-txt-recognition below is illustrative; you get the exact model name on deployment.)

$ curl -sS https://api.axforge.ai/v1/chat/completions \
  -H "Authorization: Bearer $AXFORGE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"meiki-txt-recognition","messages":[{"role":"user","content":"Hello"}]}'

Details

Languages, data & research

Languages

en ja

Tags

tensorrt onnx text-recognition textline_recognition computer-vision japanese ocr OCR image-to-text en ja

Licence

Open, with conditions

Open weights under lgpl-3.0, which carries conditions (e.g. attribution or an acceptable-use clause). Worth a read before production use — we can walk you through it. Read the licence ↗

Sources

Weights & code

Want meiki.txt.recognition on EU-owned hardware?

Request this model on EU hardware See what’s served now

Explore

More language models

© 2026 AxForge · EU-hosted AI infrastructure Pricing Docs Trust Privacy Terms