{"slug": "mistral-prices-regional-inference-at-10-premium-and-priority-tier-at-75", "title": "Mistral prices regional inference at 10% premium and Priority Tier at 75%", "summary": "Mistral has made regional inference generally available for API customers in the European Union and United States, charging a 10% premium for eligible input, output, and cached-token usage, with requests processed on infrastructure within that geography. The company also introduced a Priority Tier costing 1.75× standard pricing, offering a 99.5% uptime SLA and fallback to standard service. However, regional inference supports only stateless features, excluding Agents, Batch, and Files API, and control plane operations may still occur outside the selected region.", "body_md": "# Mistral prices regional inference at 10% premium and Priority Tier at 75%\n\n- Regional inference costs 1.1× standard pricing and covers only supported models and stateless inference features.\n[[1]](https://docs.mistral.ai/inference/regional-inference) - Priority Tier costs 1.75× standard pricing, requires account setup and model-specific limits, and can fall back to standard service.\n[[2]](https://docs.mistral.ai/inference/priority-tier) - Mistral’s control plane, including billing and usage metadata, may still be handled outside the selected geography.\n[[1]](https://docs.mistral.ai/inference/regional-inference)\n\nMistral has made regional inference generally available for API customers in the European Union and United States, charging a 10% premium for eligible input, output and cached-token usage. Requests sent to the EU or US endpoint are processed on infrastructure within that geography, while Mistral’s default global endpoint carries no location guarantee. [[1]](https://docs.mistral.ai/inference/regional-inference)\n\n## The guarantees stop short of the whole platform\n\nRegional inference supports only workloads whose models and features are available in the selected geography. Function calling is currently the only regional tool call; Agents, Batch and the Files API are unavailable on regional endpoints, and model availability varies by region. Account configuration, API keys, billing, access management and usage analytics may still be handled outside the selected inference geography. [[1]](https://docs.mistral.ai/inference/regional-inference)\n\nMistral also separates location from retention. Its Zero Data Retention program is available on paid plans for approved organizations and supported stateless APIs, including chat completions, embeddings, OCR and audio endpoints. It does not cover Agents, Batch files, Conversations, Libraries, the Files API, Vibe Work or Chat. [[3]](https://docs.mistral.ai/admin/monitor-comply/zero-data-retention)\n\n## A paid fast lane for eligible traffic\n\nMistral’s Priority Tier routes eligible requests ahead of Standard Tier traffic during high load. It costs 1.75× standard list pricing, requires account setup and custom model-specific limits, and carries a 99.5% uptime SLA. Requests using the automatic setting can fall back to Standard Tier when priority capacity or the customer’s configured limits are unavailable. [[2]](https://docs.mistral.ai/inference/priority-tier)\n\nThe design is broadly comparable to regional options from larger providers, though the details differ. Amazon Bedrock offers single-region and geographic cross-region profiles for US and EU processing, while OpenAI’s API supports US and Europe regional processing for eligible models and endpoints, with additional requirements for non-US regions. Google Cloud similarly offers regional and geography-based endpoints for supported models. [[4]](https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html)[[5]](https://platform.openai.com/docs/models/default-usage-policies-by-endpoint) [6] Mistral’s narrower feature coverage means customers using agents or file-backed workflows must examine the specific product path rather than assume that an EU endpoint provides end-to-end EU residency.\n\n[[1]](https://docs.mistral.ai/inference/regional-inference)## Companies mentioned\n\n## Further sources\n\n[[1] Mistral Docs, “Regional inference,” covering the 1.1× price, EU and US endpoint… ↗](https://docs.mistral.ai/inference/regional-inference)\n\n[[2] Mistral Docs, “Priority Tier,” covering eligibility, fallback behavior, 99.5% S… ↗](https://docs.mistral.ai/inference/priority-tier)\n\n[[3] Mistral Docs, “Zero data retention,” covering supported stateless endpoints and… ↗](https://docs.mistral.ai/admin/monitor-comply/zero-data-retention)\n\n[[4] Amazon Web Services documentation, “Regional availability by models” and “Geogr… ↗](https://docs.aws.amazon.com/bedrock/latest/userguide/models-region-compatibility.html)\n\n[[5] OpenAI Platform documentation, “Data controls in the OpenAI platform,” covering… ↗](https://platform.openai.com/docs/models/default-usage-policies-by-endpoint)\n\n[[6] Google Cloud documentation on regional and multi-region endpoints for Claude on… ↗](https://cloud.google.com/blog/products/ai-machine-learning/multi-region-endpoints-for-claude-available-on-vertex-ai)\n\nThe stories that matter, in one email. Free — unsubscribe anytime.", "url": "https://wpnews.pro/news/mistral-prices-regional-inference-at-10-premium-and-priority-tier-at-75", "canonical_source": "https://mlq.ai/news/mistral-prices-regional-inference-at-10-premium-and-priority-tier-at-75/", "published_at": "2026-08-12 13:34:56.116710+00:00", "updated_at": "2026-08-12 13:34:58.114442+00:00", "lang": "en", "topics": ["ai-products", "ai-infrastructure", "ai-policy"], "entities": ["Mistral", "Amazon Bedrock", "OpenAI", "Google Cloud"], "alternates": {"html": "https://wpnews.pro/news/mistral-prices-regional-inference-at-10-premium-and-priority-tier-at-75", "markdown": "https://wpnews.pro/news/mistral-prices-regional-inference-at-10-premium-and-priority-tier-at-75.md", "text": "https://wpnews.pro/news/mistral-prices-regional-inference-at-10-premium-and-priority-tier-at-75.txt", "jsonld": "https://wpnews.pro/news/mistral-prices-regional-inference-at-10-premium-and-priority-tier-at-75.jsonld"}}