Last verified
50% OFF TO OCT 223B ACTIVE PARAMETERS512K CONTEXTA THIRD OF SOLAR PRO 4KOREAN, ENGLISH, JAPANESE

Solar Mini 4 API Pricing

Solar Mini 4 is Upstage's compact model for agents that run at volume: a 35B-parameter mixture-of-experts that activates 3B parameters per token, with the same 512K window as Solar Pro 4. The list price is $0.10/M input, $0.01/M cached input and $0.40/M output, a third of the flagship on every line. At launch Upstage takes 50% off until October 22, 2026 (UTC): $0.05 / $0.005 / $0.20 is what an account is charged today, and list price applies from October 23. The tiles hold the list rate.

Input - per 1M tokens
$0.10/M
Charged now $0.05/M until Oct 22 list
Output - per 1M tokens
$0.40/M
Charged now $0.20/M until Oct 22 list
Cached input
$0.01/M
Charged now $0.005/M until Oct 22 -90%
Effective - agentic blend
$0.0561/M
92/8 split - 82% cache
§ 01 / TERMINAL

Run the numbers.

Calculator pre-loaded with Solar Mini 4's LIST rates - what you pay from October 23, 2026. Until October 22 the 50% launch discount applies, so halve every result for the bill you get today.

$ /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.

SUBAGENT

Tool-calling sub-task

$0.00091/call
12,000 in - 1,500 out~109,361 units/$100
SUPPORT

Korean support ticket reply

$0.00061/ticket
8,000 in - 1,000 out~164,041 units/$100
EXTRACTION

Contract field extraction

$0.00185/doc
40,000 in - 2,000 out~54,112 units/$100
LONG CONTEXT

Case file review

$0.010/review
300,000 in - 6,000 out~9,746 units/$100
§ 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 428
Words 70
Tokens (estimated) 113 tokens
Cost as input · uncached $0.00001 USD
Cost as output · uncached $0.00005 USD
Cost as cached input $0.000001 USD
§ 04 / SHELF

Up against the shelf.

All models →
Model Input /M Output /M Effective blended Context Best for
Solar Mini 4 Current $0.10 cache $0.01 $0.40 $0.0561 agentic 92/8 512K High-volume agents with a 512K window
Solar Pro 4 $0.30 cache $0.06 $1.20 $0.191 +3.4× · promo 512K The flagship, same window, three times the rate
Mistral Small 4 $0.15 cache $0.015 $0.60 $0.0842 +50% 256K Low-cost general work from a European vendor
GPT-6 Luna $0.10 cache $0.01 $0.50 $0.0682 +22% 1.05M High-volume work on a closed frontier stack
DeepSeek V4.1 Flash $0.30 cache $0.006 $1.20 $0.15 +2.7× 1M Cheap 1M-context agent loops with images
GLM-5.3-Flash $0.15 cache $0.03 $0.50 $0.0875 +56% 1M Multimodal flash tier with a 1M window

Frequently asked.

What Solar Mini 4 costs during and after its launch discount, and how it compares with the rest of the Solar line.

Q · 01 How much does Solar Mini 4 cost? +
The list price is $0.10/M input, $0.01/M cached input and $0.40/M output, $0.06/M on our 92/8 agentic blend with 82% cache hits. Until October 22, 2026 (UTC) Upstage takes 50% off, so an account is charged $0.05 / $0.005 / $0.20 today. List price applies from October 23. All Upstage prices exclude 10% VAT.
Q · 02 How does it compare with Solar Pro 4? +
Same 512K window, a third of the price: at list Solar Pro 4 costs $0.30 / $0.06 / $1.20, exactly three times Solar Mini 4 on every line. Both sit on the same rate-limit ladder. Pro 4 is Upstage's agentic flagship for document work, coding and terminal tasks; Mini 4 is the one Upstage pitches for high-frequency, cost-sensitive calls, with 3B active parameters. Both now carry the same 50% discount, so the ratio holds while they last: Pro 4 until further notice, Mini 4 through October 22.
Q · 03 Is it cheaper than Solar Pro 3? +
Yes, at list: $0.10 against $0.15 input and $0.40 against $0.60 output, with four times the context (512K against 128K). Solar Pro 3 is the larger model - 102B total and 12B active parameters - so the question is whether your task needs its capacity, not its price - and only until October 30, 2026, when Upstage ends service for Solar Pro 3.
Q · 04 What is inside it? +
Upstage's model history lists the build solar-mini4-260922 as a 35B-parameter mixture-of-experts with 3B active, a 512K context, up to 128K output tokens and a February 2026 training cutoff. It supports chat, reasoning, structured outputs and tool calling, in English, Korean and Japanese. The solar-mini4 alias currently points to that build.
Q · 05 What are the rate limits? +
They scale with your commitment tier: Tier 0 gets 100 requests and 250,000 tokens per minute, rising through 400 / 500,000 (Explore), 2,000 / 750,000 (Build) and 4,000 / 1.5M (Scale) to 8,000 / 3M on Enterprise. Accounts still on Upstage's legacy tiers are held at 100 requests and 50,000 tokens per minute on this model.
Q · 06 Is the old Solar Mini the same model? +
No. Solar Mini is Upstage's earlier 10.7B dense model with a 32,768-token context, still on the price card at a single $0.15 figure. Solar Mini 4 is a different model - a 35B mixture-of-experts with 3B active and a 512K window - and is cheaper on input at list.
Y
Reviewed by Yaroslav Vikhariev Founder · AI//COST · Prices checked by hand against upstage.ai · Last verified Sep 24, 2026
Methodology Report a correction More by Y.V.