Upstage / upstage/solar-pro3

Solar Pro 3 - access through LLMTR

Solar Pro 3 is Upstage's large-scale model, optimized for stronger reasoning and better instruction following. Its 131,072-token context window carries long documents and multi-step tool loops in a single session; there is no separate output ceiling, so generated tokens come out of that same window. Reasoning is off by default. You select a level with `reasoning_effort` across none, minimal, low, medium, and high; the model starts thinking at `medium` — none, minimal, and low are accepted but produce no reasoning tokens at all. This differs from Solar Pro 4, where the threshold is `low`. When reasoning is on, the trace is returned separately in the response's `reasoning` field and streamed incrementally. Every `tool_choice` value works for native tool/function calling, and it produces JSON object output as well as schema-conforming structured output. Repeated prompt prefixes engage the prompt-cache discount. It accepts text only; image and audio input are not supported. Official language coverage is English, Korean, and Japanese. The `xhigh` and `max` levels do not exist on this model — sending them returns an error listing the levels that do.

Technical specifications

Canonical IDupstage/solar-pro3
ProviderUpstage
Context window131,072 tokens
OperationsCHAT_COMPLETIONS
Modalitiestext

Pricing

An 8% platform margin applies to credit top-ups; model usage prices are not separately marked up.

OperationMetricUnitPrice
CHAT_COMPLETIONSINPUT_TEXTPER_1M_TOKENS$0.150000
CHAT_COMPLETIONSCACHE_READPER_1M_TOKENS$0.015000
CHAT_COMPLETIONSOUTPUT_TEXTPER_1M_TOKENS$0.600000

Example usage

With existing OpenAI SDK flows, change only the base URL and model identifier.

curl https://llmtr.com/v1/chat/completions \
  -H "Authorization: Bearer llmtr-your_key" \
  -H "Content-Type: application/json" \
  -d '{"model":"upstage/solar-pro3","messages":[{"role":"user","content":"Hello"}]}'

Related models