Solar Pro 4 Benchmarks, Specifications & Availability

Solar Pro 4

Explore Solar Pro 4 from Upstage: published specifications, source-linked vendor benchmarks, independent evaluator coverage and recorded pricing when available.

Published specifications

Provider: Upstage
Access: Proprietary
License: Proprietary
Context window: 512K
Total parameters: Not published
Active parameters: Not published
Released: 2026-08-11
Modalities: text
Family: Solar

Model card · Announcement · Website · OpenRouter

Model notes

Upstage closed commercial flagship for text input/output, English/Korean/Japanese. Official Hugging Face org checked 16 October 2026: Solar Pro 4 has a tokenizer but no released exact model weights; Solar Open2-250B is a separate model, not these weights. Total/active parameters explicitly undisclosed in vendor docs. Vendor specifies 512K context / 128K maximum output; OpenRouter serves 524,288 context / 131,072 maximum output. Announced date follows the 11 August launch article; API docs separately identify solar-pro4-260806 as released 6 August, and press release was published 20 August. Reasoning enabled; low/high supported per vendor, but evaluation effort is not stated. Old vendor-cited index 42 and AA current estimated index 28.154 are different dated/index-version observations; do not blend. Verified main Intelligence Index left unfilled because AA marks its current aggregate estimated; explicitly labeled estimated result remains reachable. Hosted API pricing uses healthy OpenRouter endpoint minima, not expired launch promotions. Vendor documentation states a February 2026 training-data cutoff.

Compare Solar Pro 4 with other models →Compare subscription plans Explore data coverage

Pricing · OpenRouter

Dated cached OpenRouter rates in USD per 1M tokens. Open the dashboard for live enhancements. Per-metric endpoint minima can refer to different providers; they are not a guaranteed combined rate from one endpoint.

Recorded pricing tiers
Tier	Input / 1M	Output / 1M	Cached input / 1M	Cache write / 1M	Date & source
Default	$0.15	$0.6	$0.03	—	2026-10-20 · OpenRouter source

Recorded pricing notes

Min-healthy endpoint minima, status0 and uptime>=98%; providers: Upstage. Per-metric minima may use different providers. Published endpoint rates are used directly; discount flags are undocumented and not applied again. Precision/provider details retained in source endpoint receipt. No new speed estimate, expired promotion or contributor tier assumed.; min-healthy endpoint minima 2026-10-16; discount flag 0.7 on Upstage (undocumented, not applied); discount flag 0.7 on Upstage (undocumented, not applied); discount flag 0.7 on Upstage (undocumented, not applied); min-healthy endpoint minima 2026-10-17; discount flag 0.7 on Upstage (undocumented, not applied); discount flag 0.7 on Upstage (undocumented, not applied); discount flag 0.7 on Upstage (undocumented, not applied); min-healthy endpoint minima 2026-10-18; discount flag 0.7 on Upstage (undocumented, not applied); discount flag 0.7 on Upstage (undocumented, not applied); discount flag 0.7 on Upstage (undocumented, not applied); min-healthy endpoint minima 2026-10-19; discount flag 0.7 on Upstage (undocumented, not applied); discount flag 0.7 on Upstage (undocumented, not applied); discount flag 0.7 on Upstage (undocumented, not applied); min-healthy endpoint minima 2026-10-20; discount flag 0.5 on Upstage (undocumented, not applied); discount flag 0.5 on Upstage (undocumented, not applied); discount flag 0.5 on Upstage (undocumented, not applied)

Official / vendor benchmarks

Default headline records. Own-vendor, peer-vendor and third-party provenance remain visible in evidence; configurations may differ.

Official / vendor headline scores; expand evidence for every record
Benchmark / evaluator	Headline score	Evidence
Terminal-Bench 2.1	57%unknown effort	All 1 recorded result & sources 57% · raw 57 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage vendor-cited Artificial Analysis August 2026 observation, not direct current AA data. The chart labels these internally measured while prose cites AA; provenance disagreement is retained and not resolved by guessing. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
GDPval-AA v2 (Upstage reported score)	38.8 scoreunknown effort	All 1 recorded result & sources 38.8 score · raw 38.8 score Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage vendor-cited Artificial Analysis August 2026 observation, not direct current AA data. 38.8 is a native reported score; quantity/normalization is unpublished, so it is not merged into GDPval Elo. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
BrowseComp	49.2%unknown effort	All 1 recorded result & sources 49.2% · raw 49.2 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
SWE-bench Verified (OpenHands)	70.6%unknown effort	All 1 recorded result & sources 70.6% · raw 70.6 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. OpenHands explicitly named; not a default unspecified agent. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
MCP Atlas	61.4%unknown effort	All 1 recorded result & sources 61.4% · raw 61.4 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
APEX-Agents	18.7%unknown effort	All 1 recorded result & sources 18.7% · raw 18.7 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
τ³-bench Banking	23%unknown effort	All 1 recorded result & sources 23% · raw 23 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage vendor-cited Artificial Analysis August 2026 observation, not direct current AA data. The chart labels these internally measured while prose cites AA; provenance disagreement is retained and not resolved by guessing. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
AA-LCR	71%unknown effort	All 2 recorded results & sources 71% · raw 71 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage vendor-cited Artificial Analysis August 2026 observation, not direct current AA data. The chart labels these internally measured while prose cites AA; provenance disagreement is retained and not resolved by guessing. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4 71% · raw 71 % Alternative · unknown effort · Own vendor Source/record date: 2026-08-20 Later vendor press release repeats August AA-LCR observation; retained as a source alternative to launch article. https://www.upstage.ai/news/upstage-ai-unveils-solar-pro-4-scoring-42-on-artificial-analysis-index-to-rank
GPQA Diamond	89%unknown effort	All 1 recorded result & sources 89% · raw 89 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage vendor-cited Artificial Analysis August 2026 observation, not direct current AA data. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
MMLU-Pro	86.3%unknown effort	All 1 recorded result & sources 86.3% · raw 86.3 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
LiveCodeBench (Upstage, revision unspecified)	87.8%unknown effort	All 1 recorded result & sources 87.8% · raw 87.8 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. LiveCodeBench revision/date window unspecified, not silently v6. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
AIME 2026	95.3%unknown effort	All 1 recorded result & sources 95.3% · raw 95.3 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
KMMLU-Pro	79.2%unknown effort	All 1 recorded result & sources 79.2% · raw 79.2 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
KBL (Korean law)	77.5%unknown effort	All 1 recorded result & sources 77.5% · raw 77.5 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
KorMedMCQA	93.2%unknown effort	All 1 recorded result & sources 93.2% · raw 93.2 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
Ko-GDPval	87.3%unknown effort	All 1 recorded result & sources 87.3% · raw 87.3 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
Arena-Hard v2 (Japanese)	81.9%unknown effort	All 1 recorded result & sources 81.9% · raw 81.9 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
MMLU (Japanese)	89.5%unknown effort	All 1 recorded result & sources 89.5% · raw 89.5 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
IFEval (Japanese)	81.6%unknown effort	All 1 recorded result & sources 81.6% · raw 81.6 % Headline · unknown effort · Own vendor Source/record date: 2026-08-11 Harness: Upstage internal evaluation, common vendor environment; exact harness/version, task subset, sampling and effort unspecified. Exact approved Solar Pro 4; Solar Open 2 is a separate, unapproved comparison peer. Checked 16 October 2026. https://www.upstage.ai/blog/en/solar-pro-4
AA Intelligence Index (Upstage August 2026 citation)	42unknown effort	All 1 recorded result & sources 42 · raw 42 index Headline · unknown effort · Own vendor Source/record date: 2026-08-20 Historical August vendor-cited AA aggregate. Current index revision differs and AA current aggregate is estimated; do not compare 42 as current v4.3.2. https://www.upstage.ai/news/upstage-ai-unveils-solar-pro-4-scoring-42-on-artificial-analysis-index-to-rank

Independent evaluators

Evaluator harnesses are distinct from vendor measurements. Missing coverage is not a failed test.

Independent evaluator headline scores; expand evidence for every record
Benchmark / evaluator	Headline score	Evidence
GDPval-AA v2.1 Elo · Artificial Analysis	1106unknown effort	All 1 recorded result & sources 1106 · raw 1106 Elo Headline · unknown effort · Independent evaluator Source/record date: 2026-10-20 Manually observed public GDPval-AA v2.1 leaderboard table, full 298-configuration pool checked 20 October 2026. Exact source configuration: Solar Pro 4; creator: Upstage; source rank 142; source release-month label Aug 2026 is model release metadata, not execution date. Published rounded native Elo 1106; literal displayed CI -20 / +20. Native Elo is not a percentage and is not the normalized (Elo−500)/2000 chart quantity. This versioned metric is separate from older generic GDPval-AA Elo and vendor GDPval/GDPval-AA results; no old evidence is overwritten. Execution date, inference precision, unprinted effort budgets and task costs are unpublished in this captured table. Elo is relative to the dated comparison pool; drift from an earlier observation does not prove a new evaluation run. Source does not label an effort; unknown is preserved rather than copying a vendor default. Headline: highest published native Elo for this exact approved model; tied values prefer existing featured effort then source order. https://artificialanalysis.ai/evaluations/gdpval-aa
Intelligence Index (estimated) · Artificial Analysis	28.2unknown effort	All 1 recorded result & sources 28.2 · raw 28.1541333537469 index Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. AA explicitly marks this aggregate estimated; do not rank as the fully measured Intelligence Index or replace vendor historical42. Changelog15Oct reports27, while current page28.1541333537469; retain this dated current observation. https://artificialanalysis.ai/models/solar-pro4
AA-Omniscience Index · Artificial Analysis	-0.8unknown effort	All 1 recorded result & sources -0.8 · raw -0.85 index Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. https://artificialanalysis.ai/models/solar-pro4
AA-Omniscience accuracy · Artificial Analysis	18.9%unknown effort	All 1 recorded result & sources 18.9% · raw 18.933333333333334 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
AA-Omniscience hallucination rate · Artificial Analysis	24.4%unknown effort	All 1 recorded result & sources 24.4% · raw 24.403782894736842 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
GDPval-AA Elo · Artificial Analysis	1107.5unknown effort	All 1 recorded result & sources 1107.5 · raw 1107.46 Elo Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. https://artificialanalysis.ai/models/solar-pro4
Terminal-Bench 2.1 (AA) · Artificial Analysis	57.3%unknown effort	All 1 recorded result & sources 57.3% · raw 57.30337078651681 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
Terminal-Bench 4.0 (AA) · Artificial Analysis	0.5%unknown effort	All 1 recorded result & sources 0.5% · raw 0.505050505050505 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
SciCode (AA) · Artificial Analysis	44.6%unknown effort	All 1 recorded result & sources 44.6% · raw 44.560185185185205 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
AA-LCR · Artificial Analysis	74%unknown effort	All 1 recorded result & sources 74% · raw 74 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
Humanity's Last Exam (AA harness) · Artificial Analysis	29.2%unknown effort	All 1 recorded result & sources 29.2% · raw 29.2400370713624 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
GPQA Diamond (AA) · Artificial Analysis	89.1%unknown effort	All 1 recorded result & sources 89.1% · raw 89.0909090909091 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
CritPt (AA) · Artificial Analysis	5.4%unknown effort	All 1 recorded result & sources 5.4% · raw 5.42857142857143 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
GDP.PDF (AA) · Artificial Analysis	5.6%unknown effort	All 1 recorded result & sources 5.6% · raw 5.6000000000000005 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
AutomationBench (AA) · Artificial Analysis	9.1%unknown effort	All 1 recorded result & sources 9.1% · raw 9.082506839361528 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. Partial reward, not strict all-assertion pass rate. https://artificialanalysis.ai/models/solar-pro4
τ³-Banking (AA) · Artificial Analysis	23.3%unknown effort	All 1 recorded result & sources 23.3% · raw 23.298969072164898 % Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. Published fraction ×100; raw fraction retained. https://artificialanalysis.ai/models/solar-pro4
Output speed · Artificial Analysis	85.7 tokens/sunknown effort	All 1 recorded result & sources 85.7 tokens/s · raw 85.6804345167237 tokens/s Headline · unknown effort · Independent evaluator Source/record date: 2026-10-16 Independent AA exact model page. Reasoning enabled; selectable effort unspecified (reasoningTokens is not an inference budget). Index v4.3.2 where stated. Inference precision and execution dates unpublished. AA source: {'type': 'firstParty', 'providerName': 'Upstage'}; API prompt/harness observation, not local inference or OpenRouter endpoint speed. https://artificialanalysis.ai/models/solar-pro4

Read how we select and source scores or the comparison guide.