Model
Model explorer

Upstage Solar Pro 2 (Chat)

CLOSED
Upstage · SOLAR family · released Jul 10, 2025

31B 'frontier' model billed as Korea's first; Chat mode is the fast default, Reasoning mode trades latency for accuracy on hard tasks.

ReasoningCodingVisionFunction callingTool useAgentic
1103.4
Elo · unrated
Parameters
31B
Active params
31B (dense)
Context
64K tokens
Architecture
Dense transformer with switchable Chat / Reasoning (chain-of-thought) inference modes
License
Proprietary
Languages
API price (in/out)
$0.15 / $0.6
Modalities
text
Benchmark results
Bar shows position within the tracked field; marker = field best
GPQA DiamondReasoning56.0%#175
best: GPT-5.6 · 94.6%
Humanity's Last ExamReasoning4.0%#135
best: Claude Sonnet 5 · 57.4%
IFBenchReasoning34.0%#61
best: MiniMax M3 · 83.0%
Run it locally
Closed weights — available via API only. No local deployment.
Input / M tok
$0.15
Output / M tok
$0.6
API price $0.15/$0.6 · each benchmark row carries its own source badge (see methodology)