PRICE~moonshotai/kimi-latest cached_input_per_mtok decreased from 0.80 to 0.292mPRICE~moonshotai/kimi-latest output_per_mtok decreased from 13.00 to 11.362mPRICE~moonshotai/kimi-latest input_per_mtok decreased from 0.99 to 0.752mPRICE~deepseek/deepseek-pro-latest cached_input_per_mtok increased from 0.0042 to 0.102mPRICE~deepseek/deepseek-pro-latest output_per_mtok increased from 0.40 to 4.202mPRICE~deepseek/deepseek-pro-latest input_per_mtok decreased from 0.13 to 0.132mPRICE~moonshotai/kimi-latest cached_input_per_mtok increased from 0.29 to 0.8033mPRICE~moonshotai/kimi-latest output_per_mtok increased from 11.36 to 13.0033mPRICE~moonshotai/kimi-latest input_per_mtok increased from 0.75 to 0.9933mPERFORMANCEGLM-5.2 → Scaleway: TTFT ↑ 20%1hPERFORMANCEDeepSeek V4 Flash → Scaleway: TTFT ↑ 35%1hPERFORMANCEMistral Medium 3.5 → Scaleway: TTFT ↓ 21%1hPERFORMANCEDeepSeek V4 Flash → Scaleway: throughput ↓ 20%1hPERFORMANCEGPT-OSS 120B → Scaleway: throughput ↑ 31%1hPERFORMANCEGPT-OSS 120B → Scaleway: TTFT ↓ 24%1hPERFORMANCEGPT-OSS 120B → Together AI: throughput ↑ 20%1hPERFORMANCEKimi K3 via OpenRouter: TTFT ↓ 49%1hPERFORMANCEGLM-5.2 via OpenRouter: reliability recovered 96.8% → 100.0%1hPERFORMANCEKimi K3 via Cortecs: throughput ↑ 184%1hPERFORMANCEKimi K3 via Cortecs: TTFT ↓ 42%1hPERFORMANCEGLM-5.2 via OpenRouter: throughput ↓ 49%1hPERFORMANCEGLM-5.2 via OpenRouter: TTFT ↑ 121%1hPERFORMANCEMiniMax M3 via OpenRouter: reliability recovered 96.8% → 100.0%1hPERFORMANCEDeepSeek V4 Pro via Cortecs: throughput ↑ 27%1hPERFORMANCEGPT-OSS 120B via OpenRouter: throughput ↓ 22%1hPERFORMANCEGLM-4.7 via OpenRouter: TTFT ↓ 78%1hPERFORMANCEGPT-OSS 20B → Groq: TTFT ↑ 24%1hPERFORMANCEKimi K3 → Together AI: TTFT ↓ 34%1hPERFORMANCEDeepSeek V4 Flash via OpenRouter: throughput ↓ 17%1hPERFORMANCELlama 3.3 70B via Cortecs: TTFT ↓ 51%1hPERFORMANCEQwen3 235B via OpenRouter: throughput ↑ 37%1hPERFORMANCEQwen3 235B via OpenRouter: TTFT ↓ 29%1hPERFORMANCEDeepSeek V4 Flash → Together AI: throughput ↑ 19%1hPERFORMANCEGLM-5.3 via OpenRouter: throughput ↓ 17%1hPERFORMANCEGPT-OSS 20B via Cortecs: throughput ↑ 28%1hPERFORMANCELlama 3.3 70B → Together AI: TTFT ↓ 33%1hPRICE~moonshotai/kimi-latest cached_input_per_mtok decreased from 0.80 to 0.291hPRICE~moonshotai/kimi-latest output_per_mtok decreased from 13.00 to 11.361hPRICE~moonshotai/kimi-latest input_per_mtok decreased from 0.99 to 0.751hPRICE~deepseek/deepseek-v4-flash-latest cached_input_per_mtok decreased from 0.0077 to 0.00111h
MARKET REPORT

The European Inference Report

What is actually measurable in European AI inference today — and the size of the gap the registry says is coming.

Key findings
  • 3 of 8 measured providers qualify as European — EU-based or offering EU data residency.
  • Mistral is the strongest measured European path at 93.2.
  • The registry catalogues far more European providers than the measured set — the coverage gap is the story, and it closes one measurement credential at a time.

The measured European ranking

Ranked under the same methodology as the global table, restricted to providers that are EU-based or offer EU data residency: Mistral (93.2), Scaleway (92.4), Cortecs (92.1). Every number is measured from Falkenstein — the vantage a European workload actually experiences.

Sovereignty is not a static attribute but a supply chain: EU headquarters, EU processing, EU-owned infrastructure. The dossier of each provider carries its classification; this ranking only admits paths that satisfy the residency bar.

The gap between catalogued and measured

The registry tracks 9+ European serving options among 18 benchmark-declared providers, and the discovery engine catalogues European sovereign clouds, national hosts and specialist providers daily. Measurement follows credentials: as European providers land in the measured set, this report re-ranks automatically.

What this means for engineers
  • European teams can compare measured EU-resident paths against the global leaders on the same scale before committing — the premium, if any, is a number, not a guess.
  • For European providers: an unmeasured catalogue entry is invisible in procurement — measurement is distribution.
Dataset

Figures recompute from the live rolling window on every visit; the snapshot and versions above are the provenance of the regime that produced them. View the current benchmark →

Cite this researchInferenceBench (2026). The European Inference Report. InferenceBench Research.
← All research
Related research
The European Inference Report · InferenceBench