Upstage model product
Solar Pro 4 is Upstage's proprietary reasoning-focused agentic LLM (text→text) for finishing multi-step commercial work: long documents, terminal tasks, and multi-turn tool use.
Updated Aug 17, 2026. Default version: Solar Pro 4
This profile uses the model's current scored version. Arena ratings and prices are shown separately.
Benchmark scores for Solar Pro 4.
Benchmark | Score | Rank | Participants | Percentile | Evidence | Evaluated |
|---|
| BenchmarkTau3 Banking | Score23.0% | Rank03 | Participants7 | Percentile66.7% | EvidenceC | Evaluated |
| BenchmarkAA-LCR | Score71.0% | Rank04 | Participants18 | Percentile82.3% | EvidenceC | Evaluated |
| BenchmarkLiveCodeBench | Score87.8% | Rank04 | Participants75 | Percentile96.0% | EvidenceC | Evaluated |
| BenchmarkAIME 2026 | Score95.3% | Rank07 | Participants21 | Percentile70.0% | EvidenceC | Evaluated |
| BenchmarkMMLU-Pro | Score86.3% | Rank13 | Participants134 | Percentile91.0% | EvidenceC | Evaluated |
| BenchmarkTerminal-Bench 2.1 | Score57.0% | Rank24 | Participants28 | Percentile14.8% | EvidenceC | Evaluated |
| BenchmarkGPQA | Score89.0% | Rank29 | Participants239 | Percentile88.2% | EvidenceC | Evaluated |
| BenchmarkBrowseComp | Score49.2% | Rank48 | Participants62 | Percentile22.9% | EvidenceC | Evaluated |
| BenchmarkSWE-Bench Verified | Score70.6% | Rank62 | Participants111 | Percentile44.5% | EvidenceC | Evaluated |
Preference and agent-evaluation results for the default version.
Arena | Category | Rank | Rating / score | Votes | Observations | Result date |
|---|
| Arenaagent tool hallucination | Categoryoverall | Rank34 | Rating / score0.0 | VotesN/A | Observations151K | Result date |
| Arenaagent bash recovery steps | Categoryoverall | Rank42 | Rating / score-0.1 | VotesN/A | Observations8.6K | Result date |
| Arenaagent task outcome explicit | Categoryoverall | Rank42 | Rating / score-0.1 | VotesN/A | Observations2.2K | Result date |
| Arenaagent | Categoryoverall | Rank44 | Rating / score-0.1 | VotesN/A | Observations165.8K | Result date |
| Arenaagent praise complaint | Categoryoverall | Rank47 | Rating / score-0.2 | VotesN/A | Observations805 | Result date |
| Arenaagent steerability | Categoryoverall | Rank47 | Rating / score-0.1 | VotesN/A | Observations3.2K | Result date |
| Arenawebdev | Categoryoverall | Rank75 | Rating / score1371.3 | Votes1,400 | ObservationsN/A | Result date |
| Arenatext factuality | Categoryoverall | Rank134 | Rating / score1397.5 | Votes2,433 | ObservationsN/A | Result date |
| Arenatext | Categoryoverall | Rank150 | Rating / score1378.0 | Votes2,433 | ObservationsN/A | Result date |
| Arenatext style control | Categoryoverall | Rank166 | Rating / score1377.7 | Votes2,433 | ObservationsN/A | Result date |
Official vendor API pricing appears first, followed by individual provider offers.
Provider | Provider model ID | Region | Input / 1M | Output / 1M | Context | Updated |
|---|
| ProviderNanoGPT | Provider model IDupstage/solar-pro4 | Regionglobal | Input / 1M$0.03 | Output / 1M$0.12 | Context524.3K | Updated |
| ProviderOpenRouter | Provider model IDupstage/solar-pro4 | Regionglobal | Input / 1M$0.03 | Output / 1M$0.12 | Context524.3K | Updated |
| ProviderLLMTR | Provider model IDupstage/solar-pro4 | Regionglobal | Input / 1M$0.03 | Output / 1M$0.12 | Context524.3K | Updated |
| ProviderUpstage | Provider model IDsolar-pro4 | Regionglobal | Input / 1M$0.30 | Output / 1M$1.2 | Context524.3K | Updated |
| ProviderKilo Gateway | Provider model IDupstage/solar-pro4 | Regionglobal | Input / 1M$0.30 | Output / 1M$1.2 | Context524.3K | Updated |
Official prices use only the vendor's configured official Provider and positive standard USD PAYG rates. Third-party offers remain explicitly labeled.
Provider-specific output speed and catalog latency for Solar Pro 4. Runtime does not affect the capability score.
No provider-specific speed or latency record is linked to the default version yet.
Output Speed is generated output tokens received per second. Catalog latency is reported separately from observed provider TTFT.
Technical details for the model's default version.
Available versions of this model. The score column identifies the version used in the overall ranking.
Version | Released | LLMBoard | Parameters | Context | Max output | Open weights | License |
|---|
| VersionSolar Pro 4 | Released | LLMBoard57.3 | ParametersN/A | Context524.3K | Max output131.1K | Open weightsNo | LicenseProprietary |
Open a comparison with the three ranked models immediately above and below this model.
Recommendations prioritize the same model type and family, then the closest LLMBoard score.
Key information about Solar Pro 4 and its available data.
Solar Pro 4 is Upstage's proprietary reasoning-focused agentic LLM (text→text) for finishing multi-step commercial work: long documents, terminal tasks, and multi-turn tool use. Supports English, Korean, and Japanese with a ~512K context window (OpenRouter listing ~524K) and up to 128K output tokens.
Reasoning effort is adjustable (high for deep analysis, low for fast chat). 3. 8 does not fit that encoding (no gdpval-aa-v2 id). 20 output per 1M tokens; launch promo is 90% off on Upstage Console and OpenRouter through September 10, 2026 (23:59 UTC) — promo not stored in the provider price row.
Also available via SolarChat, Hermes Agent, and Upstage Studio.
Data as of 2026-08-17.
Common questions about Solar Pro 4.
Solar Pro 4's default version was released on Aug 6, 2026.
Solar Pro 4's official API price is $0.30 per million input tokens and $1.2 per million output tokens via Upstage. The lowest tracked third-party offer starts at $0.03 input and $0.12 output via NanoGPT.
Solar Pro 4 was created by Upstage.
The default version has a 524.3K token context window.
No. The default version is not marked as having publicly available weights.
5 provider offerings are linked to the default version.
Nearby ranked alternatives include Muse Glimmer 30B, Grok 4.20 Beta Reasoning, DeepSeek-V3.2.