Benchmark Almanac

Solar Open 2

Solar Open 2 places in the top three on 3 benchmarks.

Upstage Released July 22nd, 2026 Family Solar Open The lab's model card Compare with others
25
Benchmarks with a score
0
Ranked first, out of five or more models
3
In the top three
Upstage Jul 2026

Solar Open 2

Open weights Mid-size

145.8 #125 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • IFBench 80% #8 /80
  • AA-LCR (Artificial Analysis Long Context Reasoning) 62.3% #21 /38
  • APEX-Agents 16.6% #12 /12

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
IFBench 80% #8 of 80 Lab model card
MMLU-Pro 86.2% #18 of 167 Lab model card
CLIcK 90.7% #1 of 4 Lab model card
KMMLU-Pro 78.4% #2 of 7 Lab model card
MultiChallenge 61% #18 of 42 Lab model card
Humanity's Last Exam 28.8% #83 of 154 Lab model card
AA-LCR (Artificial Analysis Long Context Reasoning) 62.3% #21 of 38 Lab model card
HAE-RAE Bench 73.8% #2 of 3 Lab model card

Math and science

BenchmarkScoreRankSource
AIME 2026 95.7% #8 of 38 Lab model card
GPQA Diamond 86.3% #67 of 302 Lab model card
HMMT 2026 93.9% #5 of 19 Lab model card
HRM8K 92.2% #1 of 3 Lab model card
Ko-AIME 2025 (Upstage) 97.7% #1 of 2 Lab model card

Coding

BenchmarkScoreRankSource
LiveCodeBench v6 92.4% #3 of 90 Lab model card
SWE-bench Verified 70.4% #76 of 155 Lab model card
ArtifactsBench 55.9% #3 of 5 Lab model card

Agents and tools

BenchmarkScoreRankSource
Terminal-Bench Hard 28.3% #5 of 13 Lab model card
τ³-bench Banking 19.6% #15 of 22 Lab model card
MCP Atlas 58.2% #40 of 46 Lab model card

Professional work

BenchmarkScoreRankSource
KBank-MMLU (Upstage) 80.8% #1 of 3 Lab model card
KorMedMCQA 93% #2 of 4 Lab model card
KBL 75.5% #3 of 4 Lab model card
GDPval-AA v2 1,128 #25 of 27 Lab model card
APEX-Agents 16.6% #12 of 12 Lab model card
Ko-GDPval 86.8% #2 of 2 Lab model card

Its lab's numbers put it at 145.8, number 125 of 404

The Self-Reported Index fits one score to the 14 capability results Upstage reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 143.3 to 148.2. How we compute it.

Its lab's number next to a leaderboard

Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.

Benchmark Lab says Board says Gap
Ko-GDPval 86.8% Ko-GDPval (in-house) 86.8% Leaderboard Same

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

Solar-Open2-250B model card (Hugging Face) Jul 22, 2026

Its system card

Every result in the longer documents the lab published with it.

Solar Open 2 Technical Report Jul 22, 2026

Capabilities