Model library / Teriminal-RSI
Terminal-RSI-27B is a 27.8 B Qwen-based terminal-agent model for code-related multi-step tasks requiring sandboxed execution and verification.
Parameters in total
Context length in tokens, as published
Weights at the released precision
Licence declared on the repository
Overview
Terminal-RSI-27B is a 27.8 billion-parameter Qwen-based terminal-agent model released by the Terminal-RSI project. The publisher targets repository inspection, code editing, dependency repair, scientific workflow reconstruction and other multi-step tasks that require execution in a sandboxed environment and verification of outcomes. It alternates recursive self-rewrite and policy-refine stages to turn verified trajectories into reproducible behaviour.
The checkpoint comprises 13 Safetensors files totalling about 55.56 GB (51.75 GiB). It requires 27.8 billion parameters, no expert layers, and supports a 262,144-token context window. The model is released under the Apache-2.0 licence, but weight files are not published in a standard format, so loading needs Transformers 5.17.0 or later with Qwen3_5 support.
Specification
| Repository | Teriminal-RSI/Terminal-RSI-27B |
| Publisher | Teriminal-RSI |
| Published | 21 October 2026 |
| Architecture | Qwen3_5ForConditionalGeneration |
| Model type | qwen3_5 |
| Parameters | 27,781,427,952 |
| Layers | 64 |
| Hidden size | 5,120 |
| Attention heads | 24, 4 key/value heads |
| Context length | 262,144 tokens |
| Vocabulary | 248,320 tokens |
| 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 | 51.7 GiB | Full precision as released |
| fp8 | 25.9 GiB | 8-bit, near-lossless on most models |
| int4 | 12.9 GiB | 4-bit, smallest footprint |
The cache costs 256 KiB per token at 16-bit, so the full 262,144-token context of one request needs about 64.0 GiB. Concurrency multiplies that number, not the weights.
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 | 24.7 GiB | about 100,978 tokens |
| RTX 5090 32GB | 2 | 0.9 GiB | about 3,494 tokens |
| L40S 48GB | 2 | 29.7 GiB | about 121,458 tokens |
| A100 80GB | 1 | 17.8 GiB | about 72,716 tokens |
| H100 80GB | 1 | 17.8 GiB | about 72,716 tokens |
| RTX PRO 6000 96GB | 1 | 32.2 GiB | about 131,698 tokens |
| H200 141GB | 1 | 72.7 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 Teriminal-RSI/Terminal-RSI-27B \ --max-model-len 98304 \ --gpu-memory-utilization 0.90
Questions
The weights take about 52 GiB at the released precision. That fits on one A100 80GB, which leaves roughly 18 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
Nanbeige4.3-33B is a 33.1 billion-parameter Llama-style causal model released by Nanbeige for high-capacity agentic tasks and tea…
Nanbeige4.3-33B-Base is a 33.1-billion-parameter Llama-style causal language model released by Nanbeige under Apache-2.0.
Flint is a 39.8 billion-parameter language model for coding agents, using about 2.5 billion active parameters per token.
Hy-Memory-30B-A3B is a Tencent-developed long-term-memory writer model designed to replace LLMs in Mem0, MemOS and Hy-Memory fram…
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.