# Decision 3.0 d3-nano 2B > Run Decision 3.0 d3-nano 2B by vllm-sr online or via API. Pay per run, no setup. **URL:** https://inference.sh/apps/vllm-sr/d3-nano **Last updated:** 2026-10-11 --- The 2B Decision 3.0 model from vLLM Semantic Router. Send text, images and videos with typed questions (choice, score, yes/no); get a probability for every answer. - app: `vllm-sr/d3-nano` - maker: [vllm-sr](https://inference.sh/apps/vllm-sr.md) - category: [decision](https://inference.sh/apps/category/decision.md) - tasks: [classification](https://inference.sh/apps/tag/classification.md), [routing](https://inference.sh/apps/tag/routing.md), [moderation](https://inference.sh/apps/tag/moderation.md), [vision](https://inference.sh/apps/tag/vision.md), [video](https://inference.sh/apps/tag/video.md), [open weights](https://inference.sh/apps/tag/open-weights.md) - price: GPU time, billed per second - runs via: api, javascript and python sdk, belt cli, mcp ## d3-nano API Guide ### About the 2b decision 3.0 model from vllm semantic router. send text, images and videos with typed questions (choice, score, yes/no); get a probability for every answer. ### 1. Calling the API #### Install the Client ```bash pip install inferencesh ``` #### Setup Your API Key Set `INFERENCE_API_KEY` as an environment variable. Get your key from settings → api keys. ```bash export INFERENCE_API_KEY="inf_your_key" ``` #### Run and Get Result ```python import os from inferencesh import inference client = inference(api_key=os.environ["INFERENCE_API_KEY"]) result = client.run({ "app": "vllm-sr/d3-nano@0q17m8t2", "input": { "state": "The order arrived damaged yesterday. The customer has a receipt and asks for a replacement today.", "choices": [ { "id": "choices_1", "instructions": "instructions", "options": [ { "name": "options_1" }, { "name": "options_2" } ] } ], "scores": [ { "id": "scores_1", "instructions": "instructions", "levels": [ "levels_1", "levels_2" ] } ], "nouls": [ { "id": "nouls_1", "instructions": "instructions" } ] } }) print(result["output"]) ``` #### Stream Live Updates ```python import os from inferencesh import inference client = inference(api_key=os.environ["INFERENCE_API_KEY"]) ## stream=True yields updates as they arrive for update in client.run({ "app": "vllm-sr/d3-nano@0q17m8t2", "input": { "state": "The order arrived damaged yesterday. The customer has a receipt and asks for a replacement today.", "choices": [ { "id": "choices_1", "instructions": "instructions", "options": [ { "name": "options_1" }, { "name": "options_2" } ] } ], "scores": [ { "id": "scores_1", "instructions": "instructions", "levels": [ "levels_1", "levels_2" ] } ], "nouls": [ { "id": "nouls_1", "instructions": "instructions" } ] } }, stream=True): if update.get("progress"): print(f"progress: {update['progress']}%") if update.get("output"): print(f"output: {update['output']}") ``` ### 2. Authentication The API uses API keys for authentication. See the authentication docs for detailed setup instructions. ### 3. Files #### Automatic Upload ```python ## local file paths are automatically uploaded result = client.run({ "app": "vllm-sr/d3-nano@0q17m8t2", "input": { "image": "/path/to/local/image.png", # detected & uploaded "audio": "https://example.com/audio.mp3", # url passed through } }) ``` #### Manual Upload ```python ## upload and get a hosted URL file = client.files.upload("/path/to/file.png") print(file.uri) # https://cloud.inference.sh/... ``` ### 4. Schema #### Input - **state** (any): The text to evaluate: a string, or a JSON object / array of related context (messages, records, a policy). Every question sees the same state, images and videos. May be empty when `images` or `videos` carry the content. Input over the model's token limit is rejected, never truncated. - **images** (array): Up to 8 images (PNG, JPEG or WebP) the questions are about, placed before the state. Each is read at up to 1.6 megapixels; larger images cost more tokens and time. - **videos** (array): Up to 4 videos (MP4, WebM, MOV or MKV, at most 32 MB and 5 minutes each) the questions are about, placed after the images. Read at 2 frames per second, at most 32 frames per video at up to 0.2 megapixels; all videos together take at most 16,384 tokens. - **choices** (array): Choice questions: pick one option from a set. - **scores** (array): Score questions: place the state on ordered levels. - **nouls** (array): Noul questions: probability that the answer is yes. #### Output - **choices** (object): Choice answers by question id. - **input_tokens** (integer): Input tokens, summed over the questions. Image and video tokens are included. - **model** (string): The model that answered, e.g. `d3-mini`. - **nouls** (object): Noul answers by question id. - **scores** (object): Score answers by question id. ## other decision apps - [Decision 3.0 d3 27B](https://inference.sh/apps/vllm-sr/d3.md): GPU time, billed per second - [Decision 3.0 d3-flash 9B](https://inference.sh/apps/vllm-sr/d3-flash.md): GPU time, billed per second - [Decision 3.0 d3-mini 4B](https://inference.sh/apps/vllm-sr/d3-mini.md): GPU time, billed per second - [Decision 3.0 d3-lite 0.8B](https://inference.sh/apps/vllm-sr/d3-lite.md): GPU time, billed per second - [Decision 3.0 d3-edge 0.6B](https://inference.sh/apps/vllm-sr/d3-edge.md): GPU time, billed per second ## faq ### how much does Decision 3.0 d3-nano 2B cost? GPU time, billed per second. you pay per run from your inference shell credits. no subscription is required. ### how do I run Decision 3.0 d3-nano 2B via api? call vllm-sr/d3-nano with the javascript or python sdk, the rest api, or `belt app run vllm-sr/d3-nano` from a terminal. the api reference on this page lists every input. ### who makes Decision 3.0 d3-nano 2B? Decision 3.0 d3-nano 2B is made by vllm-sr and runs on inference shell. every vllm-sr app is listed at https://inference.sh/apps/vllm-sr. ### can my agent use Decision 3.0 d3-nano 2B? yes. agents call it as a tool through the inference shell mcp server (https://api.inference.sh/mcp), the belt cli or the sdk, and pay per run like you do. ### what are the alternatives to Decision 3.0 d3-nano 2B? other decision apps on inference shell include Decision 3.0 d3 27B, Decision 3.0 d3-flash 9B, Decision 3.0 d3-mini 4B, Decision 3.0 d3-lite 0.8B. they run through the same api, so switching is a one-line change.