Model library / zgcagi
ZGCM-1 is a 7.4 billion-parameter dense language model designed for mathematical reasoning and tool-assisted search.
Parameters in total
Context length in tokens, as published
Weights at the released precision
Licence declared on the repository
Overview
ZGCM-1 is a 7.39 billion-parameter dense language model released by zgcagi. According to the authors it targets mathematical reasoning and agentic search, supporting both internal thinking and direct-response modes. The model employs a hybrid-attention architecture with gated sliding-window and global attention, and was trained with FP8 precision and Muon optimisation. It is evaluated on a range of reasoning benchmarks.
In practice the model weighs about 7.4 billion parameters stored in bfloat16 format, requiring roughly 30 GB of GPU memory. It accepts sequences up to 262 144 tokens, enabling very long contexts. The licence is MIT, allowing unrestricted use. There are no expert layers, and the architecture class is ZgcmForCausalLM, accessible via the repository with trust_remote_code enabled.
Specification
| Repository | zgcagi/ZGCM-1-7B |
| Publisher | zgcagi |
| Published | 7 September 2026 |
| Architecture | ZgcmForCausalLM |
| Model type | zgcm |
| Parameters | 7,394,832,384 |
| Layers | 32 |
| Hidden size | 4,096 |
| Attention heads | 32, 8 key/value heads |
| Context length | 262,144 tokens |
| Vocabulary | 155,136 tokens |
| Weight format | bfloat16 |
| Licence | mit |
Memory
Parameter count multiplied by the bytes each format uses. The key-value cache comes on top.
| Precision | Weights | Notes |
|---|---|---|
| bf16 | 13.8 GiB | Full precision as released |
| fp8 | 6.9 GiB | 8-bit, near-lossless on most models |
| int4 | 3.4 GiB | 4-bit, smallest footprint |
The cache costs 128 KiB per token at 16-bit, so the full 262,144-token context of one request needs about 32.0 GiB. Concurrency multiplies that number, not the weights.
That figure is an upper bound: this model uses sliding-window attention with a 128-token window on some layers, so long requests cache less than this.
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 | 1 | 5.3 GiB | about 43,631 tokens |
| RTX 5090 32GB | 1 | 12.5 GiB | about 102,613 tokens |
| L40S 48GB | 1 | 26.9 GiB | about 220,578 tokens |
| A100 80GB | 1 | 55.7 GiB | the full 262,144 tokens |
| H100 80GB | 1 | 55.7 GiB | the full 262,144 tokens |
| RTX PRO 6000 96GB | 1 | 70.1 GiB | the full 262,144 tokens |
| H200 141GB | 1 | 110.6 GiB | the full 262,144 tokens |
Serving
A vLLM launch line for the shape above. Check the model card for a runtime the publisher recommends.
vllm serve zgcagi/ZGCM-1-7B \ --max-model-len 40960 \ --gpu-memory-utilization 0.90
Questions
The weights take about 14 GiB at the released precision. That fits on one RTX 4090 24GB, which leaves roughly 5 GiB for the key-value cache.
The repository declares mit. 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
NeoHorse-1-9B is a 9 billion-parameter causal language model fine-tuned for text-based agent tasks, tool use, coding and instruct…
NeoHorse-1-4B is a 4.2B open-weight model published by TokenRhythm on Hugging Face. It uses the Qwen3_5ForCausalLM architecture w…
K2-Horizon-7B is a 9-billion-parameter decoder-only language model with a 524k token context window, released under Apache-2.0.
K2-Horizon-3.7B is a 3.7 billion-parameter dense decoder-only model with a 512 K token context, released openly by IFM.
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.