# ChindaMT-4B by iApp Technology: Open-Weight Model Source: https://savrn.com/models/chindamt-4b Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve ChindaMT-4B (5.2B parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 10.3 GB | 12.4 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 5.2 GB | 6.2 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 2.6 GB | 3.1 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 7, 2026. [ChindaMT-4B on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/chindamt-4b/gpus) ## Model Card By iApp Technology, published under apache-2.0, revision 8c1a44cd82e5. Paper: [Reference-Grounded Data Curation for Instruction-Following Thai-English Machine Translation](https://arxiv.org/abs/2609.34770), AACL-IJCNLP 2026 Main Conference ChindaMT-4B is an open-weight Thai-English machine translation model fine-tuned from [Qwen/Qwen3.5-4B](https://savrn.com/models/qwen3-5-4b) on [Grounded](https://huggingface.co/datasets/iapp/ChindaMT-Grounded), a 1.97M-record dataset built by Reference-Grounded Data Curation (RGDC). It translates in both directions and follows auxiliary rules given in the prompt, such as terminology, register, length, and output format. It is one of three sizes in the ChindaMT family (4B, 2B, 0.8B). - Task: Thai-English machine translation with instruction following - Base model: Qwen3.5-4B - Parameters: 4B - License: Apache-2.0 (inherits the base-model license) ### Prompting Plain translation. Same template for both directions; swap the language line and the source tag: ``` Translate English to Thai. EN: The weather is nice today. ``` ``` Translate Thai to English. TH: วันนี้อากาศดีมาก ``` With rules. Add a Rules: block between the language line and the source line. Rules are free-form text: [Read the full model card (949 words)](https://savrn.com/models/chindamt-4b/card) ## Configuration Architecture Qwen3_5ForConditionalGeneration Context length (tokens) 262,144 Layers 32 Hidden size 2,560 Feed-forward size 9,216 Attention heads 16 Key/value heads 4 Head dimension 256 Vocabulary size 248,320 Model type qwen3_5 ## Identity and Version Repository iapp/ChindaMT-4B Publisher iApp Technology Task Translation Modality Text Library transformers Parameters 5.2B parameters Languages en, th Revision 8c1a44cd82e595c027326c121a0548d94169d944 First published 2026-04-23 Last updated 2026-09-29 ## Files and Weights 10 files, 10.4 GB in total. The weights are 1 file totalling 10.4 GB in safetensors. Weights1 file · 10.4 GB Configuration3 files · 4.4 KB Tokenizer2 files · 20.0 MB Documentation1 file · 8.3 KB Other2 files · 943.7 KB Repository1 file · 1.6 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model.safetensors | Weights | 10.4 GB | c53667291ea7 | | config.json | Configuration | 2.9 KB | — | | generation_config.json | Configuration | 163 B | — | | processor_config.json | Configuration | 1.3 KB | — | | README.md | Documentation | 8.3 KB | — | | assets/banner.png | Other | 935.9 KB | 8350328b300a | | chat_template.jinja | Other | 7.8 KB | — | | .gitattributes | Repository | 1.6 KB | — | | tokenizer.json | Tokenizer | 20.0 MB | 87a7830d63fc | | tokenizer_config.json | Tokenizer | 1.2 KB | — | ## License and Download License apache-2.0 Access Open weights, no gate Download size 10.4 GB [Download from iApp Technology](https://huggingface.co/iapp/ChindaMT-4B) Released by iApp Technology through its official repository on Hugging Face. [Read the license](https://www.apache.org/licenses/LICENSE-2.0). ## Built From - Derived from [Qwen/Qwen3.5-4B](https://savrn.com/models/qwen3-5-4b) - Described by arXiv:2609.34770 - Trained on (disclosed) iapp/ChindaMT-Grounded ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 10.4 GB | | 16-bit | 10.3 GB | | 8-bit | 5.2 GB | | 4-bit | 2.6 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Questions About ChindaMT-4B ### How much GPU memory does ChindaMT-4B need? About 12.4 GB at 16-bit and 3.1 GB at 4-bit: the weights (5.2B parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run ChindaMT-4B on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### Can I use ChindaMT-4B commercially? Yes. ChindaMT-4B is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors. ### What is ChindaMT-4B's context length? 262,144 tokens, from the maximum position embeddings in its published configuration. ## Similar Models Model · Translation ### [madlad400-3b-mt](https://savrn.com/models/madlad400-3b-mt) [Google](https://savrn.com/model-publishers/google) MADLAD-400-3B-MT is a multilingual machine translation model based on the T5 architecture that was trained on 1 trillion tokens covering over 450 languages using publicly available data. It is competitive with models that are significantly larger. Disclaimer: Juarez Bochi, who was not involved in this research, converted the original weights and wrote the contents of this model card based on the original paper and Flan-T5. Find below some example scripts on how to use the model: First, install the Python packages that are required: pip install transformers accelerate sentencepiece Usage with candle: We also provide a quantized model (1.65 GB vs the original 11.8 GB file): See the research… Open weights apache-2.0 2.9B parameters transformers [View model](https://savrn.com/models/madlad400-3b-mt) Model · Translation ### [t5-3b](https://savrn.com/models/t5-3b) [T5 community](https://savrn.com/model-publishers/google-t5) The developers of the Text-To-Text Transfer Transformer (T5) write: T5-3B is the checkpoint with 3 billion parameters. The developers write in a blog post that the model: See the blog post and research paper for further details. The model is pre-trained on the Colossal Clean Crawled Corpus (C4), which was developed and released in the context of the same research paper as T5. The model was pre-trained on a on a multi-task mixture of unsupervised (1.) and supervised tasks (2.). Thereby, the following datasets were being used for (1.) and (2.): 1. Datasets used for Unsupervised denoising objective: 2. Datasets used for Supervised text-to-text language modeling objective - CoLA Warstadt et… Open weights apache-2.0 2.9B parameters transformers [View model](https://savrn.com/models/t5-3b) Model · Translation ### [Hy-MT2-1.8B](https://savrn.com/models/hy-mt2-1-8b) [Tencent](https://savrn.com/model-publishers/tencent) English ｜ 中文 Hy-MT2 is a family of “fast-thinking” multilingual translation models designed for complex real-world scenarios. It includes three model sizes: 1.8B, 7B, and 30B-A3B (MoE), all of which support translation among 33 languages and effectively follow translation instructions in multiple languages. For on-device deployment, AngelSlim 1.25-bit extreme quantization reduces the storage requirement of the 1.8B model to only 440 MB and improves inference speed by 1.5x. Multi-dimensional evaluations show that Hy-MT2 delivers outstanding performance across general, real-world business, domain-specific, and instruction-following translation tasks. The 7B and 30B-A3B models outperform… Open weights apache-2.0 2B parameters 262,144 tokens transformers [View model](https://savrn.com/models/hy-mt2-1-8b) Model · Translation ### [HY-MT1.5-1.8B-FP8](https://savrn.com/models/hy-mt1-5-1-8b-fp8) [Tencent](https://savrn.com/model-publishers/tencent) Hunyuan Translation Model Version 1.5 includes a 1.8B translation model, HY-MT1.5-1.8B, and a 7B translation model, HY-MT1.5-7B. Both models focus on supporting mutual translation across 33 languages and incorporating 5 ethnic and dialect variations. Among them, HY-MT1.5-7B is an upgraded version of our WMT25 championship model, optimized for explanatory translation and mixed-language scenarios, with newly added support for terminology intervention, contextual translation, and formatted translation. Despite having less than one-third the parameters of HY-MT1.5-7B, HY-MT1.5-1.8B delivers translation performance comparable to its larger counterpart, achieving both high speed and high quality.… Open weights 1.8B parameters 262,144 tokens transformers [View model](https://savrn.com/models/hy-mt1-5-1-8b-fp8) Model · Translation ### [t5-large](https://savrn.com/models/t5-large) [T5 community](https://savrn.com/model-publishers/google-t5) The developers of the Text-To-Text Transfer Transformer (T5) write: T5-Large is the checkpoint with 770 million parameters. The developers write in a blog post that the model: See the blog post and research paper for further details. The model is pre-trained on the Colossal Clean Crawled Corpus (C4), which was developed and released in the context of the same research paper as T5. The model was pre-trained on a on a multi-task mixture of unsupervised (1.) and supervised tasks (2.). Thereby, the following datasets were being used for (1.) and (2.): 1. Datasets used for Unsupervised denoising objective: 2. Datasets used for Supervised text-to-text language modeling objective - CoLA Warstadt… Open weights apache-2.0 738M parameters transformers [View model](https://savrn.com/models/t5-large) Model · Translation ### [mbart-large-50-many-to-many-mmt](https://savrn.com/models/mbart-large-50-many-to-many-mmt) [AI at Meta](https://savrn.com/model-publishers/facebook) This model is a fine-tuned checkpoint of mBART-large-50. mbart-large-50-many-to-many-mmt is fine-tuned for multilingual machine translation. It was introduced in Multilingual Translation with Extensible Multilingual Pretraining and Finetuning paper. The model can translate directly between any pair of 50 languages. To translate into a target language, the target language id is forced as the first generated token. To force the target language id as the first generated token, pass the forcedbostokenid parameter to the generate method. See the model hub to look for more fine-tuned versions. Arabic (arAR), Czech (csCZ), German (deDE), English (enXX), Spanish (esXX), Estonian (etEE), Finnish… Open weights 611M parameters 1,024 tokens transformers [View model](https://savrn.com/models/mbart-large-50-many-to-many-mmt) ## iApp Technology [All models and datasets](https://savrn.com/model-publishers/iapp) ## Versions - [8c1a44cd82e5](https://savrn.com/models/chindamt-4b/versions/8c1a44cd82e5) · current 2026-09-29 ## Explore More - [All translation models](https://savrn.com/models/tasks/translation) - [All models under apache-2.0](https://savrn.com/models/licenses/apache-2-0) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-09-29. - [Hugging Face record](https://huggingface.co/iapp/ChindaMT-4B) - [How the hub is built](https://savrn.com/model-hub/methodology)