Model library / XingChen-AGI
Xing4.0-29B-A4B is a 31.2 billion-parameter MoE language model aimed at complex engineering and agent-oriented tasks.
4 of 64 experts active per token
Context length in tokens, as published
Weights at the released precision
Licence declared on the repository
Overview
Xing4.0-29B-A4B is a mixture-of-experts model released by XingChen-AGI. The publisher describes it as built for complex engineering workloads, supporting multi-step planning, tool calling and long-context reasoning within an agent-oriented architecture. It is optimised for Ascend NPU training and intended for downstream fine-tuning on vertical domains such as intent classification, table understanding and contract auditing.
The model contains 31.2 billion parameters with 64 experts, of which four are active per token, yielding roughly 4 billion active parameters per inference step. It accepts up to 262,144 tokens of context, is licensed under Apache-2.0, and is stored in bfloat16 format. It can be served via Transformers, vLLM, SGLang, KTransformers and accessed through an OpenAI-compatible API.
Specification
| Repository | XingChen-AGI/Xing4.0-29B-A4B |
| Publisher | XingChen-AGI |
| Published | 16 September 2026 |
| Architecture | Xing4_0ForCausalLM |
| Model type | xing4_0 |
| Parameters | 31,215,031,088 |
| Experts | 64 total, 4 active per token |
| Layers | 40 |
| Hidden size | 3,584 |
| Attention heads | 32, 32 key/value heads |
| Context length | 262,144 tokens |
| Vocabulary | 131,072 tokens |
| Weight format | bfloat16 |
| Licence | apache-2.0 |
Memory
Parameter count multiplied by the bytes each format uses. The key-value cache comes on top.
| Precision | Weights | Notes |
|---|---|---|
| bf16 | 58.1 GiB | Full precision as released |
| fp8 | 29.1 GiB | 8-bit, near-lossless on most models |
| int4 | 14.5 GiB | 4-bit, smallest footprint |
The cache costs 560 KiB per token at 16-bit, so the full 262,144-token context of one request needs about 140.0 GiB. Concurrency multiplies that number, not the weights.
That figure is an upper bound: this model uses compressed key-value attention, so the real cache is a fraction of the figure above.
Hardware
At the released precision, with 90% of the card given to the server and 2.5 GiB kept for the runtime. The context column is what is left for the cache on that setup.
| GPU | Cards needed | Free for cache | Context it holds |
|---|---|---|---|
| RTX 4090 24GB | 4 | 18.3 GiB | about 34,186 tokens |
| RTX 5090 32GB | 4 | 47.1 GiB | about 88,113 tokens |
| L40S 48GB | 2 | 23.3 GiB | about 43,548 tokens |
| A100 80GB | 1 | 11.4 GiB | about 21,266 tokens |
| H100 80GB | 1 | 11.4 GiB | about 21,266 tokens |
| RTX PRO 6000 96GB | 1 | 25.8 GiB | about 48,229 tokens |
| H200 141GB | 1 | 66.3 GiB | about 124,064 tokens |
Serving
A vLLM launch line for the shape above. Check the model card for a runtime the publisher recommends.
vllm serve XingChen-AGI/Xing4.0-29B-A4B \ --max-model-len 32768 \ --gpu-memory-utilization 0.90
Questions
The weights take about 58 GiB at the released precision. That fits on one A100 80GB, which leaves roughly 11 GiB for the key-value cache.
The repository declares apache-2.0. Read the licence text before commercial use: the name alone does not tell you what is allowed.
The configuration allows 262,144 tokens. Whether the whole window is usable depends on the memory left for the cache, which the hardware table works out per card.
This page is a reference, and Aiqre does not serve this model today. We deploy open-weight models on dedicated EU hardware on request, so if you want an OpenAI-compatible endpoint for it, email [email protected] with the model name and roughly what volume you expect.
Also in the library
Pantheon-Reasoning-26B-A4B-1.1-V2 is a 26.5 billion-parameter Gemma-4 mixture-of-experts finetune aimed at role-play scenarios, i…
Nex-N2.5-mini is a 35.1 billion-parameter mixture-of-experts language model released under Apache-2.0.
LLaDA2.2-mini is a 16.3 billion-parameter MoE diffusion language model with 128 k token context, supporting Levenshtein editing.
K2-Horizon-32B-Stage1 is a 32-billion-parameter dense decoder-only model released by IFM for benchmarking agentic, coding and rea…
Hosting
Tell us the model and roughly what volume you expect. We reply with a price and an OpenAI-compatible endpoint, with your prompts never stored.