Upstageproprietary
Released: 2026-09-22Upstage: Solar Mini 4
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...
Intelligence Index
77.0/ 100
Calibrated multi-domain compositeThroughput Speed
130tok/sec
Output streaming throughputTime To First Token
280ms
Initial chunk server latencyLiveBench Contam-Free
67.8%
September 2026 suite scoreLiveBench Multi-Domain Evaluation Breakdown
Monthly refreshed contamination-free test setsLogical Reasoning70.8%
Coding & Repo Repair69.3%
Mathematics (AIME)68.5%
Data Analysis & Tables65.5%
Language & Comprehension67.0%
Instruction Following67.8%
Sourced Benchmark Evaluations (60)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| RULER | Nearing Saturation | 70.8% | 84 | 2026-09-12 | independent | Source ↗ |
| Mercury | Deprecated | 64.3% | 10 | 2026-09-12 | independent | Source ↗ |
| MGSM | Nearing Saturation | 68% | 65 | 2026-09-12 | independent | Source ↗ |
| MMLU | Saturated | 78.5% | 25 | 2026-09-12 | independent | Source ↗ |
| MMLU-Pro | Nearing Saturation | 70.8% | 75 | 2026-09-12 | independent | Source ↗ |
| MMLU-Redux | Saturated | 75.5% | 25 | 2026-09-12 | independent | Source ↗ |
| MT-Bench | Nearing Saturation | 7% | 75 | 2026-09-12 | independent | Source ↗ |
| MultiPL-E | Deprecated | 67.3% | 10 | 2026-09-12 | independent | Source ↗ |
| Needle-in-a-Haystack | Saturated | 95.4% | 44 | 2026-09-12 | independent | Source ↗ |
| OlympiadBench | Active | 58.7% | 80 | 2026-09-12 | independent | Source ↗ |
| Omni-MATH | Deprecated | 53% | 10 | 2026-09-12 | independent | Source ↗ |
| OpenAI-MRCR | Active | 64.7% | 80 | 2026-09-12 | independent | Source ↗ |
| PutnamBench | Active | 37.2% | 90 | 2026-09-12 | independent | Source ↗ |
| QASPER | Nearing Saturation | 63.9% | 65 | 2026-09-12 | independent | Source ↗ |
| QuALITY | Nearing Saturation | 67.8% | 78 | 2026-09-12 | independent | Source ↗ |
| MBPP+ | Nearing Saturation | 74.6% | 65 | 2026-09-12 | independent | Source ↗ |
| SCROLLS | Nearing Saturation | 61.6% | 65 | 2026-09-12 | independent | Source ↗ |
| SimpleQA | Active | 40% | 90 | 2026-09-12 | independent | Source ↗ |
| Soohak | Active | 54.4% | 98 | 2026-09-12 | independent | Source ↗ |
| StructFlowBench | Nearing Saturation | 57.6% | 75 | 2026-09-12 | independent | Source ↗ |
| SuperGLUE | Saturated | 76.2% | 25 | 2026-09-12 | independent | Source ↗ |
| SWE-bench | Deprecated | 49.7% | 10 | 2026-09-12 | independent | Source ↗ |
| SWE-bench Pro | Active | 45.3% | 90 | 2026-09-12 | independent | Source ↗ |
| SWE-bench Verified | Saturated | 57% | 25 | 2026-09-12 | independent | Source ↗ |
| SWE-Lancer | Deprecated | 49.7% | 20 | 2026-09-12 | independent | Source ↗ |
| TabMWP | Saturated | 65.9% | 25 | 2026-09-12 | independent | Source ↗ |
| Terminal-Bench | Active | 51.2% | 80 | 2026-09-12 | independent | Source ↗ |
| TruthfulQA | Saturated | 67.8% | 25 | 2026-09-12 | independent | Source ↗ |
| WinoGrande | Saturated | 77.7% | 28 | 2026-09-12 | independent | Source ↗ |
| ZeroSCROLLS | Nearing Saturation | 61.6% | 78 | 2026-09-12 | independent | Source ↗ |
| GLUE | Saturated | 80.7% | 28 | 2026-09-12 | independent | Source ↗ |
| AGIEval | Saturated | 64.2% | 25 | 2026-09-12 | independent | Source ↗ |
| AIME | Nearing Saturation | 68% | 75 | 2026-09-12 | independent | Source ↗ |
| ARC-AGI | Saturated | 70.2% | 25 | 2026-09-12 | independent | Source ↗ |
| ARC-AGI-2 | Nearing Saturation | 58.3% | 75 | 2026-09-12 | independent | Source ↗ |
| ARC-AGI-3 | Active | 48.6% | 93 | 2026-09-12 | independent | Source ↗ |
| BIG-Bench Hard | Saturated | 73.2% | 25 | 2026-09-12 | independent | Source ↗ |
| CodeContests | Deprecated | 47.5% | 10 | 2026-09-12 | independent | Source ↗ |
| CodeXGLUE | Deprecated | 68.7% | 23 | 2026-09-12 | independent | Source ↗ |
| CommonsenseQA | Saturated | 78.4% | 28 | 2026-09-12 | independent | Source ↗ |
| ComplexBench | Nearing Saturation | 55.6% | 80 | 2026-09-12 | independent | Source ↗ |
| CyberSecEval | Deprecated | 65.8% | 10 | 2026-09-12 | independent | Source ↗ |
| FireBench | Active | 57% | 95 | 2026-09-12 | independent | Source ↗ |
| FrontierCode | Active | 42.4% | 98 | 2026-09-12 | independent | Source ↗ |
| FrontierMath | Active | 20% | 98 | 2026-09-12 | independent | Source ↗ |
| AgentIF | Nearing Saturation | 46.1% | 80 | 2026-09-12 | independent | Source ↗ |
| GPQA | Nearing Saturation | 67.8% | 78 | 2026-09-12 | independent | Source ↗ |
| GSM-Hard | Deprecated | 68.7% | 10 | 2026-09-12 | independent | Source ↗ |
| GSM8K | Saturated | 77.3% | 25 | 2026-09-12 | independent | Source ↗ |
| HumanEval | Saturated | 83.3% | 25 | 2026-09-12 | independent | Source ↗ |
| HumanEval+ | Saturated | 78.9% | 25 | 2026-09-12 | independent | Source ↗ |
| Humanity's Last Exam | Active | 32.9% | 93 | 2026-09-12 | independent | Source ↗ |
| IFEval | Deprecated | 66.4% | 25 | 2026-09-12 | independent | Source ↗ |
| InfiniteBench | Nearing Saturation | 65.5% | 65 | 2026-09-12 | independent | Source ↗ |
| LiveCodeBench | Saturated | 59.9% | 39 | 2026-09-12 | independent | Source ↗ |
| LongBench | Nearing Saturation | 66.2% | 65 | 2026-09-12 | independent | Source ↗ |
| LongBench-Pro | Active | 63.1% | 80 | 2026-09-12 | independent | Source ↗ |
| MATH | Saturated | 70.2% | 25 | 2026-09-12 | independent | Source ↗ |
| MathArena | Active | 57.3% | 99 | 2026-09-12 | independent | Source ↗ |
| MBPP | Saturated | 77.5% | 25 | 2026-09-12 | independent | Source ↗ |
OpenAI Compatible API ExecutionAPI Ready
from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="upstage-solar-mini-4",
messages=[{"role": "user", "content": "Evaluate multi-step logic problem"}],
temperature=0.2
)
print(response.choices[0].message.content)Evaluation & Sourcing Notes
Scores listed for Upstage: Solar Mini 4 represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.