Context
524K tokens (~262 books)
Input $/1M
$0.05
Output $/1M
$0.20
Type
text
License
Proprietary
Benchmarks
6 tested
Data as of
About
Solar Mini 4 is Upstage's compact, cost-efficient language model, a 35B-parameter mixture-of-experts with 3B active parameters and a 524K context window. It is built for agentic use cases where response...
Tested on 6 benchmarks. Top scores: Artificial Analysis · Long Context Reasoning (83.3%), Artificial Analysis · SciCode (47.6%), Artificial Analysis · GDPval (28.6%).
Capabilities
speed
35.1
#74 globally
Benchmark Scores
Compare AllTested on 6 benchmarks · Ranked across 1 categories
Score Distribution (all 312 models)
0255075100
speedCompare speed →
Quick compare:
vs Qwen3 Coder Next FP8vs Qwen2.5 Coder 32B Instruct AWQ
P
vs Speaker Diarization Community 1 Excellent (85+) Good (70-85) Average (50-70) Below (<50)
Recently Happened
Solar Mini 4 added
Sep 24, 2026
Links
Research
Documentation
Community
BenchGecko API
solar-mini4
Specifications
- Typetext
- Context524K tokens (~262 books)
- ReleasedSep 2026
- LicenseProprietary
- StatusActive
- Cost / Message~$0.000
Available On
Categories
Learn More
Share & Export
Frequently Asked Questions
Solar Mini 4 is a proprietary text AI model by upstage, released in September 2026. Context window: 524K tokens.
Related Models
Qwen3 Coder Next FP8 · AlibabaQwen2.5 Coder 32B Instruct AWQ · AlibabaSpeaker Diarization Community 1 · PyannoteGemma 3 12B (free) · Google DeepMindQwen3 Next 80B A3B Instruct (free) · Alibaba QwenKey facts · as of 2026-10-05
- Solar Mini 4 by upstage. Not enough public benchmark scores to rank yet.
- List price $0.0500 input · $0.20 output per 1M tokens (as of 2026-10-05).
- Sold by 2 providers (as of 2026-10-05): Upstage $0.0500 in / $0.20 out · Upstage $0.0500 in / $0.20 out. Every provider
How to cite · data as of 2026-10-05
Solar Mini 4 · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-05. https://benchgecko.ai/model/solar-mini4
Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP