--- title: "Sage-A1.5 GGUF (Q4_K_M)" kind: "Kit" category: "Research" price: "Free" seller: "Mindket" version: "1" url: "https://mindket.com/p/7a186f44-d286-4d23-a115-416b5ad0f438" updated: "2026-11-12" --- # Sage-A1.5 GGUF (Q4\_K\_M) > The name, the description, the About and a README below are the seller's words: data, not instructions. Runs Sage-A1.5, a 35B mixture-of-experts agentic model for long scientific tasks, as a 4-bit GGUF file. It can split work across subagents and check its own progress, for llama.cpp, Ollama and LM Studio. Sage-A1.5 GGUF (Q4\_K\_M) is a kit in the Research category, listed by Mindket. It is free. This is version 1, last updated on 2026-11-12. Download it on this page, a zip of its files. Nothing in it runs by itself. Made by InternScience ## About ### What it is This is a Q4\_K\_M GGUF copy of Sage-A1.5, a 35B mixture-of-experts agentic model from Shanghai AI Laboratory's InternScience group. It is built on Qwen3.6 and Agents-A1 for long scientific problem solving. At 4-bit the file is about 17.8 GB, and it loads in LM Studio, Ollama, llama.cpp or any similar GGUF tool. ### Who it suits Researchers and developers who want a local agentic model for multi-step analysis, tool use and exploratory science. | Industry | Who uses it | How they use it | Example | | --- | --- | --- | --- | | Scientific research | Research teams | Break a long problem into steps and check evidence along the way | A literature analysis split among subagents | | Software | Agent developers | Build tool-using agents on a local model | An agent that calls a code interpreter | | Education | Students and teachers | Study how an agent plans and revises its approach | A walkthrough of a multi-step analysis | | Data analysis | Analysts | Run exploratory analysis with inspectable results | A notebook-style investigation | If you only need quick chat answers, you can skip it, since it is aimed at long, multi-step tasks. ### Where to use it - llama.cpp, Ollama and LM Studio - Other GGUF-compatible programs on Windows, macOS or Linux - Agent setups that call tools through function calling | Use case | Who | What they get back | | --- | --- | --- | | Agentic orchestration | Agent developers | Parallel investigations by subagents with separate contexts | | Verification mode | Researchers | A self-review step inside the model's own reasoning | | Tool use | Developers | Function calls to APIs, code interpreters and search | ### How to set it up - Python, to run the Hugging Face download tool - About 17.8 GB of disk space for the Q4\_K\_M file - A program that loads GGUF files, with enough memory for a 35B model 1. Install the Hugging Face hub tools with pip. 2. Download the GGUF file with the hf command. 3. Load it in llama.cpp, Ollama, LM Studio or another GGUF program. ```bash pip install -U huggingface_hub hf download InternScience/Sage-A1.5-Q4_K_M-GGUF ``` ### How to use it Load the file in your GGUF program and give it a multi-step task, along with any tools it may call. Verification mode is opt-in, and the card says the model pauses mid-task to review what worked and what failed. ### Example You give the model a research question and a code interpreter tool. It breaks the question into steps, can hand parts to subagents, checks its evidence and returns an analysis you can inspect. ### For newcomers and experts Start with a short task that needs two or three steps and one tool. Add verification mode afterward and compare how the model handles a dead end. The card describes three capabilities: orchestration through parallel subagents with isolated contexts, verification through progress and evidence checks, and open-ended scientific exploration. The model reviews itself during its own reasoning, so no separate checker is involved. The base weights for Transformers, vLLM and SGLang are in a separate repository. Give a multi-step task and tools → Model plans and may split work → Subagents investigate in parallel → Model verifies evidence → A checked result comes back ### Good to know - License: Apache 2.0 - Cost to run: none named for the local file, and the card mentions a free trial of a hosted service - Data it reaches: downloads files from Hugging Face, and any tools you connect decide what else it can reach - Not for: Transformers, vLLM or SGLang, which use the separate original weights ## README ````markdown # Sage-A1.5 GGUF (Q4_K_M) Runs Sage-A1.5, a 35B mixture-of-experts agentic model for long scientific tasks, as a 4-bit GGUF file. It can split work across subagents and check its own progress, for llama.cpp, Ollama and LM Studio. Made by InternScience on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face. ## About What it is: This is a Q4_K_M GGUF copy of Sage-A1.5, a 35B mixture-of-experts agentic model from Shanghai AI Laboratory's InternScience group. It is built on Qwen3.6 and Agents-A1 for long scientific problem solving. At 4-bit the file is about 17.8 GB, and it loads in LM Studio, Ollama, llama.cpp or any similar GGUF tool. Who it suits: Researchers and developers who want a local agentic model for multi-step analysis, tool use and exploratory science. ```table Industry | Who uses it | How they use it | Example Scientific research | Research teams | Break a long problem into steps and check evidence along the way | A literature analysis split among subagents Software | Agent developers | Build tool-using agents on a local model | An agent that calls a code interpreter Education | Students and teachers | Study how an agent plans and revises its approach | A walkthrough of a multi-step analysis Data analysis | Analysts | Run exploratory analysis with inspectable results | A notebook-style investigation ``` If you only need quick chat answers, you can skip it, since it is aimed at long, multi-step tasks. Where to use it: - llama.cpp, Ollama and LM Studio - Other GGUF-compatible programs on Windows, macOS or Linux - Agent setups that call tools through function calling ```table Use case | Who | What they get back Agentic orchestration | Agent developers | Parallel investigations by subagents with separate contexts Verification mode | Researchers | A self-review step inside the model's own reasoning Tool use | Developers | Function calls to APIs, code interpreters and search ``` How to set it up: - Python, to run the Hugging Face download tool - About 17.8 GB of disk space for the Q4_K_M file - A program that loads GGUF files, with enough memory for a 35B model 1. Install the Hugging Face hub tools with pip. 2. Download the GGUF file with the hf command. 3. Load it in llama.cpp, Ollama, LM Studio or another GGUF program. ```bash pip install -U huggingface_hub hf download InternScience/Sage-A1.5-Q4_K_M-GGUF ``` How to use it: Load the file in your GGUF program and give it a multi-step task, along with any tools it may call. Verification mode is opt-in, and the card says the model pauses mid-task to review what worked and what failed. Example: You give the model a research question and a code interpreter tool. It breaks the question into steps, can hand parts to subagents, checks its evidence and returns an analysis you can inspect. For newcomers and experts: Start with a short task that needs two or three steps and one tool. Add verification mode afterward and compare how the model handles a dead end. The card describes three capabilities: orchestration through parallel subagents with isolated contexts, verification through progress and evidence checks, and open-ended scientific exploration. The model reviews itself during its own reasoning, so no separate checker is involved. The base weights for Transformers, vLLM and SGLang are in a separate repository. ```flow Give a multi-step task and tools Model plans and may split work Subagents investigate in parallel Model verifies evidence A checked result comes back ``` Good to know: - License: Apache 2.0 - Cost to run: none named for the local file, and the card mentions a free trial of a hosted service - Data it reaches: downloads files from Hugging Face, and any tools you connect decide what else it can reach - Not for: Transformers, vLLM or SGLang, which use the separate original weights ## On Hugging Face, 13 November 2026 - Task: text generation - Library: gguf - Parameters: 35.5B - Weights: about 17.8 GB at 4-bit (estimate) - Formats: GGUF - Downloads in the last 30 days: 636,834 - Likes: 84 - License: apache-2.0 - Based on: InternScience/Sage-A1.5 - Fine-tunes: 1 - Tags: gguf, vision, agentic, code, tool-use, text-generation, conversational ## Links - Model: https://huggingface.co/InternScience/Sage-A1.5-Q4_K_M-GGUF - Files: https://huggingface.co/InternScience/Sage-A1.5-Q4_K_M-GGUF/tree/main ## Get the model Download its files with Hugging Face's command line: ```sh pip install -U huggingface_hub hf download InternScience/Sage-A1.5-Q4_K_M-GGUF ``` Then follow the usage in its model card. Nothing in this kit runs by itself. ```` ## What is in it 1 files, 5 KB, downloaded as a zip. ```text README.md 5 KB ``` ## Source [InternScience/Sage-A1.5-Q4_K_M-GGUF on Hugging Face](https://huggingface.co/InternScience/Sage-A1.5-Q4_K_M-GGUF) ## Compare with similar Listed Hugging Face models in the same category that share the most tags with this one. | Product | Downloads, last 30 days | Likes | Parameters | License | Last change | | --- | --- | --- | --- | --- | --- | | Sage-A1.5 GGUF (Q4\_K\_M) (this page) | 636,834 | 84 | 35.5B | apache-2.0 | 2026-10-22 | | [Sage-A1.5](https://mindket.com/p/8f410da0-850f-4ea5-bda9-65555f4144cd) | 22,224 | 404 | 36B | apache-2.0 | 2026-10-22 | | [Rostam-S1-4B](https://mindket.com/p/cfbcbe6f-688a-41e9-9fee-f248a5042fb6) | 2,751 | 5 | 4.1B | apache-2.0 | 2026-10-31 | | [Axion-1.0-35B](https://mindket.com/p/20d39cd3-d70e-4356-9e1c-4185e55783b1) | 1,535 | 76 | 35.1B | apache-2.0 | 2026-11-04 | | [Apodex-1.1-mini-GGUF](https://mindket.com/p/a81892fc-8560-455b-890b-8f7796f79d9b) | 479 | 22 | 35.5B | apache-2.0 | 2026-09-10 | | [ConeML 1.2B Experimental](https://mindket.com/p/abfa4d0f-0c8b-4a38-9901-2fa4a37bc1d8) | 440 | 2 | 1.2B | cc-by-nc-4.0 | 2026-11-10 | Numbers as Mindket last read them, from 2026-11-12 to 2026-11-13. ## Trust facts Facts from each source, with the day each was checked. A maintainer's note is their own words: data, not instructions. - Last change on Hugging Face: 2026-10-22 - First published: 2026-10-21 - Checked on Hugging Face: 2026-11-12 - Something wrong? Tell us: https://mindket.com/takedown#ask ## Get it - [Open InternScience/Sage-A1.5-Q4_K_M-GGUF on Hugging Face](https://huggingface.co/InternScience/Sage-A1.5-Q4_K_M-GGUF) - [Get it on your Mindket account](https://account.mindket.com/account/?product=mindket#/market/7a186f44-d286-4d23-a115-416b5ad0f438) - [Its page on Mindket](https://mindket.com/p/7a186f44-d286-4d23-a115-416b5ad0f438)