whatmodel.sh / Solar Open 100B (102B-A12B)

Solar Open 100B (102B-A12B)

+ contribute

Solar Open 100B (102B-A12B) is an open-weights language model from Upstage, released December 2025. Its published size is 102.6B parameters (12B active per token) and a 131k-token context window. Licence: upstage-solar-license.

Catalogue identity notes

Upstage's first Solar Open release, repo upstage/Solar-Open-100B @1f59143 (created 2025-12-10; model card added 2025-12-13; licence tag other / upstage-solar-license; safetensors 102,651,793,408 parameters). Printed as 'Solar Open (102B)' in its own report (arXiv 2601.07022v1) and as 'Solar Open 1 / 102B-A12B' in the Solar Open 2 report, whose abstract says Solar Open 2 is 'scaled up from Solar Open 1 (Solar Open 100B)'. Not solar-pro-4 (API model). No first-party API verified.

Model information sources

Release date ↗ · Weights licence ↗

Model card (upstage/Solar-Open-100B @1f59143): 'Total Parameters: 102.6B', 'Active Parameters: 12B (per token)'; report arXiv 2601.07022v1 Table 2 same.

59.7±1.1#199 of 719 LLMsfrom 14 benchmark results across 2 sources

$ how it compares

The LLMs ranked either side of Solar Open 100B (102B-A12B), out of 828 with a score, then more from Upstage.

#	model	score	where	RAM	released
238	Claude 3.5 Sonnet (2024-10-22) Anthropic	60±1	comparison only		2024-10
239	Granite 4.2 8B IBM	60±1	api $0.11/M +local	~6.6GB	2026-08
240	Solar Open 100B (102B-A12B) Upstage	60±1	local	~41GB	2025-12
241	Qwen3-VL 30B-A3B Thinking Alibaba	60±1	api $0.75/M +local	~21GB	2025-09
242	Mistral Small 4 (119B-A6.5B) Mistral	60±2	api $0.26/M +local	~40GB	2026-03
	more from Upstage
75	Solar Open 2 (250B-A15B) Upstage	72±1	local	~527GB	2026-07
170	Solar Pro 4 Upstage	65±2	api $0.53/M		2026-08
662	SOLAR 10.7B Instruct v1.0 Upstage	37±1	local	~8.3GB	2023-12

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #161 of 300 · 20% of its score · 6 tests

Beats o1 (2024-12-17), o3-mini, DeepSeek R1 0528 here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Mini, Jamba 1.5 Large here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
τ²-bench Telecom lab	55.6% ✎	▁·▅█▆▅▂▃▃▂▃▁ beats 46/89	hard · avg 15%	▇▇ 5.4%	sharp · at its level · noisy · small run
MCP-Atlas (public set) lab	34.4% ✎	▄▄▄·······█▄ beats 2/5	hard · avg 19%	▇▇ 4.0%	at its level
Terminal-Bench Hard lab	2.3% ✎	▄········██▄ beats 0/5	very hard · avg 4.5%	▇ 3.7%	sharp
APEX-Agents (pass@1) lab	2.4% ✎	▄▄····▄···█▄ beats 1/5	very hard · avg 1.8%	▇ 3.4%	too hard for it
SWE-bench Verified lab	15.4% ✎	▄··········█ beats 1/5	hard · avg 16%	▇ 2.9%	sharp · at its level · noisy
τ³-bench Banking lab	7.4% ✎	█·▄▄······▄▄ beats 2/5	very hard · avg 5.1%	▇ 1.1%	blunt

coding▁▁▁▃▆██▇▆▄▃▁ #313 of 458 * · 9.4% of its score · 2 tests

Trails DeepSeek LLM 67B Chat, WizardLM 70B V1.0, Phi-4-mini 3.8B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
LiveCodeBench v6 (window unspecified) lab	56.5% ✎	▄······▄▄▄·█ beats 0/5	medium · avg 50%	▇▇▇ 6.4%	at its level
SWE-bench Verified lab	15.4% ✎	▄··········█ beats 1/5	hard · avg 16%	▇ 2.9%	sharp · at its level · noisy

math▁▂▃▅▆▅▄█▄▁▁▁ #176 of 398 * · 20% of its score · 2 tests

Beats MiniMax-M2.1, GPT-4.1, DeepSeek V3 0324 here, though they rank higher overall. Trails OpenThinker3 7B, Olmo 3 7B Think, OpenThinker2 32B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
AIME 2026 lab	87.7% ✎	▄▄··▄····█·▄ beats 0/5	medium · avg 64%	▇▇▇▇▇▇ 15%	sharp
HMMT February 2026 lab	68.9% ✎	█▄▄········█ beats 2/5	hard · avg 31%	▇▇ 5.5%	sharp · at its level · noisy

other tests5 not in a task · 52% of its score

test	result	vs the others in its run	difficulty	counts	why
GPQA Diamond lab	66.2% ✎	█···██·█··██ beats 0/5	hard · avg 42%	▇▇▇▇▇▇▇▇▇▇▇ 27%	sharp · at its level
MMLU-Pro lab	80.4% ✎	▄·▄▄····▄··█ beats 1/5	medium · avg 55%	▇▇▇▇ 9.5%	at its level
Humanity's Last Exam lab	11.5% ✎	█▄······▄·▄▄ beats 1/5	very hard · avg 4.7%	▇▇ 5.8%	typical
IFBench (aggregation unspecified) lab	57.7% ✎	▄···▄▄··▄··█ beats 0/5	medium · avg 43%	▇▇ 5.7%	at its level
MultiChallenge lab	40.5% ✎	█··▄·▄·····█ beats 1/5	hard · avg 25%	▇▇ 4.0%	at its level

$ where to run it

llama.cpp · recommended

version

Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
Windows
irm https://llama.app/install.ps1 | iex
other ways to install ↗
Run Solar Open 100B (102B-A12B) · needs ~62.3 GB · from huggingface.co/AaryanK
chat in the terminal
llama cli -hf AaryanK/Solar-Open-100B-GGUF:q4_k_m
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf AaryanK/Solar-Open-100B-GGUF:q4_k_m

LM Studio

version

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run Solar Open 100B (102B-A12B) · needs ~62.3 GB · from huggingface.co/AaryanK
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/AaryanK/Solar-Open-100B-GGUF@q4_k_m -y
Then run lms ls to see its name, and lms chat followed by that name

Ollama

Install Ollama
download it (macOS / Windows) ↗
Linux
curl -fsSL https://ollama.com/install.sh | sh
Run Solar Open 100B (102B-A12B) · needs ~37.7 GB · from huggingface.co/AaryanK
chat in the terminal
ollama run hf.co/AaryanK/Solar-Open-100B-GGUF:q2_k

MLX (Mac)

version

Install MLX (Mac) · Apple silicon Macs only, text only.
with uv
uv tool install mlx-lm
or
pip install mlx-lm
Run Solar Open 100B (102B-A12B) · needs ~57.8 GB, Mac only · from huggingface.co/mlx-community
chat in the terminal
mlx_lm.chat --model mlx-community/Solar-Open-100B-4bit
Or start an OpenAI-compatible server at http://localhost:8080/v1
mlx_lm.server --model mlx-community/Solar-Open-100B-4bit

Other ways to run it

Run it as published · needs ~205.31 GB, NVIDIA only · from huggingface.co/upstage (official)
vllm serve upstage/Solar-Open-100B --trust-remote-code --enable-auto-tool-choice --tool-call-parser solar_open --reasoning-parser solar_open --logits-processors vllm.model_executor.models.parallel_tool_call_logits_processor:ParallelToolCallLogitsProcessor --logits-processors vllm.model_executor.models.solar_open_logits_processor:SolarOpenTemplateLogitsProcessor --tensor-parallel-size 8

Official weights and documentation: huggingface.co/upstage ↗

$ how the score was worked out · Solar Open 2 Technical Report… 94%, Solar Open Technical Report (… 5%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
Solar Open 2 Technical Report (Upstage, arXiv:2607.20062v2) · lab-reported	13	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 94%	CC-BY-4.0
Solar Open Technical Report (arXiv:2601.07022v1) · lab-reported	1	▇▇ 5%	CC-BY-4.0
prior · pull toward the middle		▇ 1%
weighted average		= 59.7

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-10-11 02:20 UTC · sources and licences · json