Compare AIFind AIAI NewsAI How-To
About Us
PrivacyTermsFAQContactContact
AIB Inc.Company info
© 2026 AIB Inc.
Upstage
Upstage

Solar Mini 4

Compare Now
Compare
Release Date 2026-09-23(Knowledge cutoff 2026-02)··524K IN · 131K OUTProprietary Model|API available
Overall2.5Top 47%
AutomationResearching
Office3.5Top 14%Coding2.5Top 50%

Solar Mini 4 is Upstage's compact language model, built for agentic workloads where speed and cost matter most. It uses a mixture-of-experts architecture. The model is large overall, but it draws only on the parts it needs for each response, which keeps quality high while cutting serving costs substantially. It can take in long documents in a single pass and produce lengthy outputs. It also supports a step-by-step reasoning mode, tool calling, and structured outputs that follow a defined schema. It is fluent in Korean and strong in English and Japanese as well.

On paper, it closely resembles its larger sibling, Solar Pro 4, with the same document length limits, the same languages, and the same training cutoff. The two differ in how demanding a task they can handle. Pro 4 is the one to reach for when an agent has to plan on its own, juggle tools, and recover from its own mistakes across many steps. Mini 4 costs far less, which makes it the efficient choice for high-volume, repetitive work such as extraction and classification. Send everyday bulk processing to Mini 4 and save the demanding agentic tasks for Pro 4 to get the best of both cost and performance.

  1. 01ImagesGenerated images
  2. 02CostCost · Usage · Speed
  3. 03User CompareUser Comparisons
  4. 04Expert EvalExpert Performance Evaluation
  5. 05OpinionPublic Opinion
  6. 06Related Info

API cost

Provider price
$0.05IN$0.2OUT
Calculate Cost
SourceOpenRouterLast updatedSep 24, 2026

Real-world usage

No.3213 places0.5% share · 336B tokens/wk
384B
9/28
336B
10/5
SourceOpenRouterLast updatedOct 8, 2026

Output Speed

Slowper 2,000 tokens
First Output1.5s
total44.9s
46tok/s
First Output 1.5stotal: 44.9s
SourceOpenRouterLast updatedOct 8, 2026

Expert Performance Evaluation

Overall performance

MoreAll overall performance AI rankings
QHow strong is it across many tests combined?AA Intelligence Index·Bottom 45%#66
24.1

About AA Intelligence IndexMore about AA Intelligence Index

Documents and office work

MoreAll documents and office work AI rankings
QCan it pull scattered facts from several long documents into one answer?Long Context Reasoning·Top 12%#13
83.3%

About Long Context ReasoningMore about Long Context Reasoning

Coding

MoreAll coding AI rankings
1.0%47.6%

About Terminal-Bench 4.0More about Terminal-Bench 4.0

Math and science

MoreAll math and science AI rankings
QCan it solve problems that stump experts across fields?Humanity's Last Exam·Bottom 36%#76
25.8%

About Humanity's Last ExamMore about Humanity's Last Exam

See each evaluation’s detail page for its source and last update·Data sources & removal requests

Related Info

Upstage
PrivateAI Native
Upstage
🇰🇷 Yongin2020~$1B
Solar DecideSolar Pro 4Solar Pro 3