# Tele-AI/Telechat

Repository: https://github.com/Tele-AI/Telechat
Canonical: https://ross.abutalabs.com/products/telechat
Language: Python
License Family: other
Last push: 2026-07-09T05:43:11+00:00

## Health v2 (maintenance only)
Score: 67/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 91, release rhythm 35, longevity 69
- inputs: {"age_days": 969, "days_push": 55, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1853, forks 100 (observed 2026-08-28T04:05:44.534581+00:00)

## What it is
TeleChat is a family of open-source bilingual (Chinese-English) large language models (1B, 7B, 12B) developed by China Telecom's AI team, released with Hugging Face-format weights, int8/int4 quantized versions, and a 1T-token Chinese training dataset. The repository provides inference, deployment (including vLLM), fine-tuning, and quantization tooling for the models.

## Use cases
- run a chinese-english chat llm locally
- deploy a 7B or 12B language model with vllm
- fine-tune an open-source chinese llm
- quantize a large language model to int8 or int4
- download pretrained chinese llm weights for huggingface
- build a chatbot on an open chinese base model
- serve an llm on domestic chinese hardware

## When to choose
- you need a strong open-weight Chinese-language chat model
- you want multiple model sizes (1B/7B/12B) with quantized variants
- you need vLLM-based serving or fine-tuning recipes for the model
- you want the accompanying large-scale Chinese training dataset

## When to avoid
- you need a permissive license - no license is specified in the repo
- you need a frontier multilingual model rather than Chinese/English focus
- you want a managed API rather than self-hosted weights

## Facets
- artifact type: library
- maturity: active
- function: llm-inference, llm-training, machine-learning, deep-learning
- domain: large-language-models, artificial-intelligence, machine-learning
- platform: python
- tags: llm, chinese-language-model, model-weights, quantization, vllm, huggingface, chat-model, decoder-only, telechat, natural-language-processing, gpu, linux, docker

## Member repositories
- Tele-AI/Telechat (main) score 67

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:44.534581+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:16:50.006976+00:00, confidence not recorded.
  - readme: https://github.com/Tele-AI/Telechat (fetched 2026-08-28T04:05:44.534581+00:00, sha d1571f64e9ce)
- Data as of 2026-08-30T08:39:29.467469+00:00.