models/upstage/solar-pro-3
U
upstage·active

Upstage: Solar Pro 3

upstage's efficiency model. Context window: 131K tokens.

Overall score
3.54
/5.00 · ranked #143
Input
$0.150
per 1M tokens
Output
$0.600
per 1M tokens
Context
131K
tokens
Blended
$0.487
3:1 out:in ratio

Price drops, new benchmarks, model updates. Stay current on Upstage: Solar Pro 3.

One email per change. Unsubscribe anytime.

modelpicker.aipowered by live benchmark data

Scores by test

Methodology →
Structured Output
5.0
Strategic Analysis
4.0
Constrained Rewriting
4.0
Creative Problem Solving
3.0
Tool Calling
4.0
Faithfulness
4.0
Classification
3.0
Long Context
4.0
Safety Calibration
2.0
Persona Consistency
4.0
Agentic Planning
3.0
Multilingual
5.0
Tabular Data
1.0
SciCode
24.7

What you need to know

Solar Pro 3 is optimized for high-precision formatting and global accessibility, achieving perfect internal scores for structured output and multilingual capabilities. Its strength in strategic analysis and constrained rewriting makes it a reliable choice for developers needing strict adherence to schema or translation-heavy workflows.

The model is priced competitively at a blended cost of $0.487 per million tokens, offering a low-cost entry point for high-context tasks given its 131K window. However, this value is offset by significant performance gaps in specific data types; it is nearly incapable of handling tabular data and shows a tendency to over-refuse benign requests due to poor safety calibration.

Technical performance is inconsistent across complex reasoning tasks. While it handles tool calling and faithfulness well, it struggles with scientific coding, as evidenced by a 24.7% SciCode score, and shows mediocre results in agentic planning and creative problem solving.

Use this model if your primary requirement is reliable JSON/structured output or multilingual support at a low price point. Skip this model if your application requires tabular data processing, high-level scientific coding, or a nuanced safety filter that avoids over-refusal.

Strengths — Top 3

Structured Output5.0/5.0
Multilingual5.0/5.0
Strategic Analysis4.0/5.0

Relative weaknesses — Bottom 3

Tabular Data1.0/5.0
Safety Calibration2.0/5.0
Creative Problem Solving3.0/5.0

Similar models

OOpenAI: gpt-oss-20b$0.1053.62IIBM: Granite 4.1 8B$0.0873.54MMinistral 3 8B 2512$0.1503.54MDevstral 2 2512$1.604.00