Last verified
DISCOUNT LADDERAGENTIC FLAGSHIPREASONS BY DEFAULTOPENAI-COMPATIBLEKOREA - UPSTAGE

Solar Pro 4 API Pricing

Solar Pro 4 is Upstage's agentic flagship, built for tool calling, terminal tasks and long-document reasoning. The list price is $0.30/M input, $0.06/M cached input and $1.20/M output, and the launch discount on top of it comes down in steps rather than ending. The step in effect now is 50% off until further notice — $0.15/M input, $0.03/M cached, $0.60/M output — which started on October 11, 2026, the day Upstage's card had promised list price. Before it came 90% off to September 10 and 70% off from September 11 to October 10; each step was announced as the last. The tiles below hold the list rate, because it is the one figure in this schedule that has never moved.

Input - per 1M tokens
$0.30/M
Charged now $0.15/M, 50% off list
Output - per 1M tokens
$1.20/M
Charged now $0.60/M, 50% off list
Cached input
$0.06/M
Charged now $0.03/M, 50% off -80%
Effective - agentic blend
$0.191/M
92/8 split - 82% cache
§ 01 / TERMINAL

Run the numbers.

Calculator pre-loaded with Solar Pro 4's LIST rates — what you pay once the discount ends. Upstage currently bills 50% off until further notice, so halve every result for the bill you get today. Solar Pro 4 reasons by default and its reasoning tokens are billed as output, so a task that thinks hard costs more than its visible answer suggests.

$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
Open full calculator (all models · share URL · CSV) →
§ 02 / SCENARIOS

Real-world presets.

§ 03 / TOKENIZER

Paste text. See tokens. See cost.

What's a token →
Estimate · solar-tokenizer-estimate · ≈3.8 chars/token Auto-counts as you type

This is a chars-per-token approximation, not a real tokenizer. Actual tokens vary by language, code density, and tool-call overhead — counts are typically ±10–20% off for English prose, more for code or non-Latin scripts. For exact billing, use the vendor's official tokenizer.

Characters 574
Words 95
Tokens (estimated) 151 tokens
Cost as input · uncached $0.00005 USD
Cost as output · uncached $0.00018 USD
Cost as cached input $0.00001 USD
§ 04 / SHELF

Up against the shelf.

All models →
Model Input /M Output /M Effective blended Context Best for
Solar Pro 4 Current $0.30 cache $0.06 $1.20 $0.191 agentic 92/8 512K Tool calling and terminal work
Solar Mini 4 $0.10 cache $0.01 $0.40 $0.0561 −3.4× · promo 512K Same 512K window at a third of the rate
Solar Pro 3 $0.15 cache $0.015 $0.60 $0.0842 deprecated · −2.3× 128K Previous mid tier, retires Oct 30
GLM-4.7 $0.60 cache $0.11 $2.20 $0.358 +88% 200K Comparable Chinese coding model

Frequently asked.

Solar Pro 4 pricing questions, with the list price kept apart from the discount schedule running on top of it.

Q · 01 How much does Solar Pro 4 cost? +
List price is $0.30/M input, $0.06/M cached input and $1.20/M output. The launch discount on top of it has come down in three steps: 90% off billed $0.03 / $0.006 / $0.12 through September 10, 2026 (UTC), 70% off billed $0.09 / $0.018 / $0.36 through October 10, and since October 11 50% off billing $0.15 / $0.03 / $0.60 until further notice. Upstage publishes no end date for the current step.
Q · 02 Why do you show the higher price? +
Because it is the one you will still be paying when the discounts run out, and a budget built on a temporary rate breaks the moment it lapses. Solar Pro 4 is the argument for that policy: the discount was published as ending on September 11, then quietly became a second phase at 70% off, whose end date moved twice before it closed on October 10 — and the list price promised for October 11 became a third phase at 50% off with no end date at all. The list price did not move once. Our catalogue stores list prices throughout so models stay comparable, and the discounted figure is printed beside every tile rather than hidden.
Q · 03 Are reasoning tokens billed separately? +
No, but they are billed. Upstage's docs state Solar Pro 4 "performs reasoning by default" and count reasoning tokens inside usage.completion_tokens, which means they bill at the output rate. Check usage.completion_tokens_details.reasoning_tokens to see how much of a bill is thinking rather than answer.
Q · 04 What context window does it have? +
512K, with up to 128K output tokens. Upstage's model history page lists the current build, solar-pro4-260806, with a 512K context length and a February 2026 training cutoff. We could not find the figure on any first-party page when this model was added on August 16, 2026, and the docs page we checked then no longer exists, so we cannot say whether it was published at the time.
Q · 05 Is the API hard to migrate to? +
It is OpenAI-compatible. Point the OpenAI SDK at https://api.upstage.ai/v1 and pass solar-pro4 as the model. The alias currently resolves to the dated build solar-pro4-260806. Rate limits scale with your commitment tier, from 100 requests and 250,000 tokens per minute on Tier 0 to 8,000 and 3M on Enterprise; accounts on Upstage's legacy tiers stay at 100 requests and 50,000 tokens per minute.