How does pay-as-you-go billing work?

You top up credits once and use them across every model. Each request is charged at the listed price per million tokens, per image or per minute of audio. There is no subscription and no minimum spend.

Should I choose a plan or pay-as-you-go?

A plan suits steady, daily coding work: a fixed monthly price for a set of coding models. Pay-as-you-go suits variable or production API traffic, and gives you every model in the catalogue at its list price.

What is cached input?

When a request repeats a prompt prefix the model has already processed, such as a long system prompt or a codebase, those tokens are read from the prompt cache. They are billed at the lower Cache In price. Models without a Cache In price bill all input at the regular rate.

How accurate is the cost calculator?

It multiplies your token numbers by the current list price of the model you pick, so the maths is exact. Your real bill depends on how many tokens your requests actually use and how often the cache is hit.

Do prices include VAT?

No. All prices are net, plus statutory VAT. Our offer is intended exclusively for businesses as defined by § 14 BGB (German Civil Code).

Pricing — Melious

Plans

Choose a subscription that fits your coding workflow. Upgrade or cancel anytime.

Basic

For lightweight Coding workflows

99 € € /mo

>2× more tokens than Claude's 99 € plan*

Drop-in for Cursor, Codex, Claude Code and more
Agents allowed
EU-hosted
No lock-in — cancel monthly

Current models (Claude: ≈250M tokens/month*)

DeepSeek V4.1 Flash 1017
up to 3B tokens/month
GLM 5.3
up to 700M tokens/month
GLM 5.3 Flash
up to 6.5B tokens/month

Start Coding

Best value

Pro

For your daily coding workloads

199 € € /mo

>2× more tokens than Claude's 199 € plan*

Drop-in for Cursor, Codex, Claude Code and more
Agents allowed
EU-hosted
No lock-in — cancel monthly

Current models (Claude: ≈1B tokens/month*)

DeepSeek V4.1 Flash 1017
up to 10B tokens/month
GLM 5.3
up to 2.5B tokens/month
GLM 5.3 Flash
up to 25B tokens/month

Start Coding

* Estimate. Compared with Claude Max 5x (€99) and Max 20x (€199) as of Q4 2026. Anthropic doesn't publish token limits, and token sizes differ between models.

This offer is intended exclusively for businesses as defined by § 14 BGB (German Civil Code) · all prices plus statutory VAT.

Or pay as you go

Pay-as-you-go

Access 60+ open-weight models on a pay-as-you-go basis. No commitments, no minimums. Just top up credits and go.

Large Language Models

Top 10 of 29, ranked as in the Hub

Model	Input (/ 1M tokens)	Output (/ 1M tokens)	Cache In (/ 1M tokens)	Actions
DeepSeek V4.1 Flash	€0.2	€1	€0.01	View Model
DeepSeek V4 Flash 0731	€0.1	€0.25	€0.02	View Model
GLM 5.3 Flash	€0.1	€0.4	€0.02	View Model
GLM 5.3	€1	€3	€0.2	View Model
Gemma 4 31B	€0.1	€0.3	€0.02	View Model
Kimi K3	€2.85	€14.25	€0.75	View Model
Qwen3.8 27B	€0.4	€2.4	€0.1	View Model
Kimi K2.6	€0.55	€3.3	€0.25	View Model
DeepSeek V4.1 Flash 1017	€0.3	€1.2	€0.03	View Model
Qwen 3.6 27B	€0.45	€3.15	—	View Model

DeepSeek V4.1 Flash

Input €0.2

Output €1

Cache In €0.01

DeepSeek V4 Flash 0731

Input €0.1

Output €0.25

Cache In €0.02

GLM 5.3 Flash

Input €0.1

Output €0.4

Cache In €0.02

GLM 5.3

Input €1

Output €3

Cache In €0.2

Gemma 4 31B

Input €0.1

Output €0.3

Cache In €0.02

Kimi K3

Input €2.85

Output €14.25

Cache In €0.75

Qwen3.8 27B

Input €0.4

Output €2.4

Cache In €0.1

Kimi K2.6

Input €0.55

Output €3.3

Cache In €0.25

DeepSeek V4.1 Flash 1017

Input €0.3

Output €1.2

Cache In €0.03

Qwen 3.6 27B

Input €0.45

Output €3.15

Explore all models in the Hub

Estimate your cost

Pick a model and your expected usage. The estimate uses current list prices.

Estimated cost

€2.95

Input 5 M × €0.2: €1.00
Cached input 95 M × €0.01: €0.95
Output 1 M × €1: €1.00

An estimate, not a quote. You pay for the tokens you actually use.

What our customers say

We put Melious through a thorough hands-on test. With GLM 5.2 we measured more than 350 tokens per second, and a development task that used to take 45 minutes was done in about 7. It stayed stable throughout, entirely on European infrastructure, and our data is not used for training. That convinced us to integrate Melious as a provider in ALBERT | AI, the AI in agorum core pro.

Oliver Schulze CEO & CTO of agorum

If inference stalls, the mule stumbles. Running agents smoothly only works when the inference keeps up. With Melious, we can be sure it does.

Bernd Paulus CTO & Co-Founder of mulios

Melious isn’t just a supplier for us, they’re responsive, reliable, and always thinking along with us.

Robert Keus CEO of GreenPT

Pricing questions

You top up credits once and use them across every model. Each request is charged at the listed price per million tokens, per image or per minute of audio. There is no subscription and no minimum spend.
A plan suits steady, daily coding work: a fixed monthly price for a set of coding models. Pay-as-you-go suits variable or production API traffic, and gives you every model in the catalogue at its list price.
When a request repeats a prompt prefix the model has already processed, such as a long system prompt or a codebase, those tokens are read from the prompt cache. They are billed at the lower Cache In price. Models without a Cache In price bill all input at the regular rate.
It multiplies your token numbers by the current list price of the model you pick, so the maths is exact. Your real bill depends on how many tokens your requests actually use and how often the cache is hit.
No. All prices are net, plus statutory VAT. Our offer is intended exclusively for businesses as defined by § 14 BGB (German Civil Code).

Top up and start building.

No subscription, no minimum. Your credits work with every model on this page.

Top Up Credits Talk to Sales