Solar-Open-100B-NVFP4A16

This repository contains a build of upstage/Solar-Open-100B that FuriosaAI quantized in-house to NVFP4A16, together with a Furiosa Executable Bundle (FXB) for running it on FuriosaAI RNGD with Furiosa-LLM. The NVFP4A16 weights in this repository also run directly on other frameworks (such as vLLM, SGLang, and Transformers); for the base model and general usage, see the upstream upstage/Solar-Open-100B model card.

Overview

Solar-Open-100B is a large-scale open-weight language model developed by Upstage. It is an auto-regressive Mixture-of-Experts (MoE) transformer that supports English and Korean, and it handles both reasoning and non-reasoning chat as well as tool (function) calling. Its intended use is the same as the upstream upstage/Solar-Open-100B, and it is released under the Upstage Solar License.

• Architecture: SolarOpen (Mixture-of-Experts)
• Input / Output: Text / Text
• Supported Inference Engine: Furiosa LLM
• Supported Hardware: FuriosaAI RNGD

Quantization

The weights are quantized to NVFP4 (4-bit floating point), while activations and the KV cache remain in 16-bit precision (NVFP4A16).

Features

• Reasoning. Solar-Open is a reasoning model. Launch the server with --reasoning-parser solar_open to have the chain of thought returned in a separate field. The reasoning depth is controllable per request through reasoning_effort ("low", "medium", "high"; default "high").
• Tool calling. The model supports tool (function) calling through the solar_open tool-call parser.

Parallelism Strategy

On RNGD, Solar-Open-100B-NVFP4A16 runs with a tensor-parallel size of 32 PEs, which maps to four RNGD cards (8 PEs per card).

Usage

To run this model with Furiosa-LLM, follow the example commands below after installing Furiosa-LLM and its prerequisites.

Launch the server

Serve the model with the solar_open reasoning parser so the chain of thought is returned in a separate field:

furiosa-llm serve furiosa-ai/Solar-Open-100B-NVFP4A16 \
  --reasoning-parser solar_open

To also enable tool (function) calling, add the solar_open tool-call parser; keep --reasoning-parser solar_open so thinking is still parsed into its own field:

furiosa-llm serve furiosa-ai/Solar-Open-100B-NVFP4A16 \
  --reasoning-parser solar_open \
  --enable-auto-tool-choice \
  --tool-call-parser solar_open

When the server is ready, you will see:

INFO:     Started server process [27507]
INFO:     Waiting for application startup.
INFO:     Application startup complete.
INFO:     Uvicorn running on http://0.0.0.0:8000 (Press CTRL+C to quit)

Basic Usage

The server exposes an OpenAI-compatible API. You can send a request with curl:

curl http://localhost:8000/v1/chat/completions \
    -H "Content-Type: application/json" \
    -d '{
    "model": "furiosa-ai/Solar-Open-100B-NVFP4A16",
    "messages": [{"role": "user", "content": "What is the capital of France?"}]
    }' \
    | python -m json.tool

With --reasoning-parser solar_open, Solar-Open returns its reasoning separately from the final answer:

• response.choices[].message.reasoning (non-streaming)
• response.choices[].delta.reasoning (streaming)

from openai import OpenAI

client = OpenAI(base_url="http://localhost:8000/v1", api_key="EMPTY")

response = client.chat.completions.create(
    model="furiosa-ai/Solar-Open-100B-NVFP4A16",
    messages=[{"role": "user", "content": "How many r's are in 'strawberry'?"}],
)

print("Reasoning:", response.choices[0].message.reasoning)
print("Answer:", response.choices[0].message.content)

Note: The reasoning field is not part of the OpenAI API specification but is a widely followed convention (the OpenAI Agents SDK, vLLM, and others). It appears only in responses that contain reasoning content; accessing it otherwise raises an AttributeError.

Advanced Usage

Controlling reasoning effort. You can control how much the model reasons per request with the reasoning_effort parameter ("low", "medium", or "high"; default "high"). Solar-Open's chat template maps this to its thinking behavior: "high" and "medium" produce a chain of thought — with "high" allowed a larger reasoning-token budget than "medium" — while "low" (and "minimal") prefill an empty thinking block so the model skips reasoning and answers directly:

response = client.chat.completions.create(
    model="furiosa-ai/Solar-Open-100B-NVFP4A16",
    messages=[{"role": "user", "content": "How many r's are in 'strawberry'?"}],
    extra_body={"reasoning_effort": "high"},
)

print("Reasoning:", response.choices[0].message.reasoning)
print("Answer:", response.choices[0].message.content)

Tool calling. With the server launched using --enable-auto-tool-choice --tool-call-parser solar_open (see Launch the server), pass tools in the request and let the model decide when to call them. See the Tool Calling guide for a complete client example and details on tool-choice options.

Learn more

• Tool Calling — parsers, tool-choice options, and more examples
• Furiosa-LLM Server (furiosa-llm serve) — full OpenAI-compatible API reference and serving options
• upstage/Solar-Open-100B — upstream model card