European LLM price index

At a glance

Findings

13 of the 20 models have an approved European endpoint. Claude Fable 5.1, GPT-6 Luna, Gemini 4 Argon, Gemini 3.8 Flash, DeepSeek V4 Flash, MiMo V2.6 Flash and Qwen 3.8 27B have none yet.

Frontier models cost more on their European cloud endpoints: 10% to 20% above the maker's own price (Claude Opus 5.5 +10%, GPT-6.1 Terra +20%, GPT-5.6 Luna +10%).

GLM 5.3 Flash, GLM 5.2 and GLM 5.3 are served in Europe within 5% of the maker's own price.

European hosting costs more for MiniMax M3 (+60%) and DeepSeek V4 Pro 0813 (+127%).

The cheapest model in the index to run on a European endpoint is GLM 5.3 Flash, at $0.175 / $0.450 on Inceptron.

Frontier models

Open models

How Standard Compute uses these prices

Standard Compute is one API for frontier and open models. It routes each step to the cheapest model that can do it and the cheapest provider fast enough to serve it, measured on latency and tokens per second. The provider prices in this index are the inputs to that choice.

Set a Europe rule for your company, a team or a single key, and every request, including retries and fallbacks, is served only by approved European inference endpoints, the rows marked Yes above. If no European endpoint can serve it, the request is refused before your prompt is sent. Our EU gateway terminates TLS and runs in Finland, and your account data is stored in Frankfurt.

Create a company workspaceor write to contact@standardcompute.com

Method

Each price is a provider's list rate in US dollars per million tokens, at the base tier: no prompt caching, long-context surcharges or tool fees. Our model catalog records every provider's rate once a day, leaves out providers reporting a degraded endpoint, and this index is republished from that record. This edition uses the record of November 9, 2026.

A row is marked Yes only when that endpoint is on the reviewed list a Europe rule routes to. For the European endpoints of Microsoft Azure, Google Vertex AI, Amazon Bedrock and Mistral, the rate is that endpoint's own European quote from our regional price list (last reviewed October 27, 2026). Providers that serve only from Europe are shown at their listed rate. Where an endpoint is approved but this edition has no European rate for it, the table names it without a price.

The models are chosen from the most requested models on Standard Compute. Each table is sorted by input plus output rate, and cheapest means the lowest of that sum. A table shows every European endpoint, the maker's own listing and the five cheapest other providers; the full list for every model is in the dataset. Providers list these rates in US dollars, so the index does too. Cite as: Standard Compute, European LLM price index, November 9, 2026.

Questions

Related

• Standard Compute pricing
• LLM API prices from each model maker
• Standard Compute for companies
• Sub-processors and European inference providers
• Security review pack