Upstage: Solar Pro 4

Upstage's flagship model, specialized for agentic use

solar-pro4

AvailableUpstageChat

Input / Output: $0.30 / $1.20
Context: 512K
Max output: 128K
Modalities: Chat
Capabilities: Tool usePrompt Cache

About Solar Pro 4

Solar Pro 4 is Upstage's flagship language model, announced in August 2026 and built for agent work that spans several documents, terminal tasks, and long chains of tool calls. It reads and writes English, Korean, and Japanese, takes text only, and has a reasoning mode whose effort can be set high for depth or low for real-time replies.

Where it works well

Multi-turn tool calling over a large knowledge base, which Upstage measures with the τ³-Banking benchmark.
Terminal and long-document work: Upstage reports 57 on Terminal-Bench v2.1 and 71 on AA-LCR for long-document reasoning.
English, Korean, and Japanese are covered for both input and output.
Reasoning effort is a dial, so one model serves both deep analysis and low-latency chat.

When to choose another model

It is a text model and does not take image input, so screenshots, scanned pages, and photos need a separate vision model.
Upstage states support for English, Korean, and Japanese only, so output quality in other languages is untested by the maker.

Getting started

Create API key
Create a key in Console, then use it with every model on the platform.
02
Send your first request
Copy the example for your language and run it against the endpoint.
Text to textResponses API
POST/v1/responses
API format:
curl https://api.tokenlab.sh/v1/responses \ -H "Content-Type: application/json" \ -H "Authorization: Bearer sk-xxx" \ -d '{ "model": "solar-pro4", "input": "Hello!" }'

Pricing

The Verified price applies to Verified, which costs less on most models. The Official price is the model maker's published price and applies to the more reliable Official route. Auto bills the route that completes the request.

Default spec

per 1M tokens

Official price: Input $0.30 / Output $1.20 / Cache read $0.06
Verified price: —
Discount: —

Prompt cache pricing

Cache read

Official price: $0.06
Verified price: —
Discount: —

	Official priceper 1M tokens	Verified priceper 1M tokens	Discount
Default spec	Input $0.30 / Output $1.20 / Cache read $0.06	—	—
Prompt cache pricing
Cache read	$0.06	—	—

Usage & activity

Success rate is the share of requests that completed. Latency is how long a full response takes; P95 means 95% of requests finished within that time.

Usage & availability

Last 24 hours

Requests
Success rate
P95 latency
Total tokens

Model performance

Last updated Nov 10, 2026, 11:49 PM UTC

Metrics appear once privacy and data volume thresholds are met.

Data is based on aggregate user requests, excluding status checks.

Open in Console

Open Solar Pro 4 in Console with a prompt ready to edit or send.

Help me try solar-pro4 with a short message at /v1/responses. Show the reply, latency, and cost.

Use cases

Multi-document agents
Give it a folder of reports and a set of tools, and have it gather facts across files, run the checks, and produce the finished deliverable instead of a partial draft.
Terminal automation
Let it work in a shell to inspect logs, run scripts, and fix configuration, stopping to report when the evidence runs out.
Korean and Japanese support bots
Answer customer policy questions in Korean or Japanese by looking up the matching rule in a knowledge base and acting on it through tool calls.

Prompt examples

Read these three vendor contracts and the pricing spreadsheet, then list every clause that conflicts with our procurement policy and say which document it comes from.

You have a shell and a log directory. Find why the nightly export failed, fix the cause, rerun it, and tell me what you changed.

고객이 환불 규정을 문의했습니다. 지식 베이스에서 해당 정책을 찾아 적용하고 처리 결과를 알려 주세요.

This model has conditional pricing. Monthly token totals alone cannot produce a reliable estimate; use the detailed pricing for the request specification, cache, and applicable time window.

FAQ

What is Solar Pro 4 designed for?

Upstage designed it for commercial agent workloads: jobs that move across several documents, run tasks in a terminal, and chain tool calls over many steps. It is meant to carry the work to a finished result and to say so when the evidence is not enough to finish.

Does Solar Pro 4 have a reasoning mode?

Yes. Reasoning is part of the model, and Upstage describes reasoning effort as a dial: high for deep analysis, low for real-time interaction. Pick low for chat-style turns and high for planning, debugging, or multi-document analysis.

Which languages does Solar Pro 4 support?

Upstage states English, Korean, and Japanese for both input and output. If your users write in other languages, test with your own samples first, since the maker does not list them.

Can Solar Pro 4 read images?

No. Upstage presents it as a text model, so it takes text only. For screenshots or scanned documents, run OCR or a vision model first and pass the extracted text to Solar Pro 4.

Does Solar Pro 4 work with existing OpenAI-style code?

Yes. Upstage describes it as OpenAI-compatible, so an existing OpenAI-style client works after you change the base URL and set the model name to solar-pro4. Anthropic-style clients can call it too.

How much does Solar Pro 4 cost?

On TokenLab, Solar Pro 4 costs Input $0.30 / Output $1.20 per 1M tokens. The pricing table above shows the full breakdown. Rates depend on the billing unit, specification, and usage. Compare matching conditions in the model's detailed pricing; a single rate does not determine the total cost.

What are the context window and output limit of Solar Pro 4?

Solar Pro 4 accepts up to 524,288 tokens of context and returns up to 131,072 tokens in one response.

Which endpoint should Solar Pro 4 use?

Use https://api.tokenlab.sh/v1/responses for Solar Pro 4. The request example below shows the matching code shape.

Which operations does Solar Pro 4 support?

Solar Pro 4 supports Text to text. Select an operation above to see its endpoint and request example.