China Telecom Open-Sources Xing4.0-29B-A4B MoE Model
11/07/2026 — 11/07, 04:15·1 sources (all)·1 reports (all)
Key takeaways
- China Telecom AI Technology has open-sourced the Xing4.0-29B-A4B MoE model on Hugging Face. Featuring 29B total and 4B active parameters wit…
Background & analysis
On November 7, 2026, a post appeared on Reddit's LocalLLaMA community showcasing Xing4.0-29B-A4B and linking to its model card on Hugging Face. According to the post, the model is part of the Xing series (formerly TeleChat) developed by China Telecom Artificial Intelligence Technology Co., Ltd., and has been released as an open-weight model hosted on Hugging Face.
The model card describes Xing4.0-29B-A4B as a Mixture-of-Experts (MoE) large language model with a total parameter count of 29B, of which only 4B are activated per token. It natively supports a context length of 256K and can be extended to 512K. The card also states that it is the first model of this scale trained entirely on Ascend architecture, marking the migration of the whole training pipeline to domestic hardware.
At this point, the Reddit post and the Hugging Face model card are the only sources available, so no independent benchmark results, deployment details, or technical reports have surfaced yet. The release itself appears to have been shared quietly within the community rather than announced through an official press release, which leaves room for further testing and discussion on how the model performs against other recent MoE implementations.
AI-generated from 1 reports · updated 1 hour ago
- Heat index
- 94
- All sources
- 1
- All reports
- 1
- First seen
- 2 hours ago
Reports on this story headlines open the original
The Xing4.0-29B-A4B open-source MoE model developed by China Telecom Artificial Intelligence Technology Co., Ltd. is now available on Hugging Face. With 29B total parameters (4B activated), it natively supports a 256K context length and is the first model in the series trained entirely on Ascend hardware.
Reddit · LocalLLaMAAI score 63
Other stories people are talking about
- Heat index 856RisingUS tech stocks rebound as investors reset Anthropic IPO valuation expectations to about $1.1 trillion12 sources
- Heat index 833OpenAI rolls out GPT-6 Cyber to Daybreak Red users with API, Codex access16 sources
- Heat index 755RisingThinking Machines Lab releases Inkling 2 Small amid reported $5B funding round11 sources
- Heat index 709SurgeMeta adds Spaces and a new model to Muse in the US10 sources
- Heat index 695Claude rolls out Personal Finance to US Pro, Max and Ultra users via read-only Plaid links13 sources
How is heat calculated?About the methodHide
Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.
This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.
- Surge
- Discussion rising fast
- New
- First report within 6 hours
- Rising
- Still gathering discussion