Solar Pro 2 (Preview) Non-reasoning is Upstage's latest flagship in preview, posting an intelligence index of 10 alongside MMLU-Pro of 0.725 and GPQA of 0.544. LiveCodeBench at 0.385 is respectable for a non-reasoning variant, suggesting usable coding capability for lighter engineering tasks.

For businesses, it fits general content generation, document analysis, and moderate coding assistance. The preview status means it may not yet be production-ready for mission-critical workflows, and the absence of pricing data makes cost planning difficult. There are no agentic or tool-use benchmarks, limiting its appeal for autonomous agent pipelines.

Teams evaluating Upstage's Solar family for content and general reasoning tasks will find this a capable mid-tier option, but should wait for a stable release and confirmed pricing before committing to production use.

Assessed June 30, 2026

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?