Cursor vs. STACKIT AI: data protection per plan
Cursor (5 plans) and STACKIT AI (1 plan) side by side: DPA, training on inputs, hosting and data retention per plan, based on the providers’ public documents. Without rating.
Last updated: (latest review date of the tools shown)
At a glance
Each value is listed with the plans it applies to. Pick a tier to see only plans of that tier; if plans are chosen in the plan comparison below, the table shows exactly those.
All plans: each value is listed with the plans it applies to.
| Feature | Cursor | STACKIT AI |
|---|---|---|
| Provider based in | Anysphere, Inc., 2261 Market Street STE 86466, San Francisco, CA 94114, USA | Schwarz Digits Cloud GmbH & Co. KG, Am Campus 1, 74177 Bad Friedrichshall, Germany |
| Plans recorded | 5 | 1 |
| Hosting | mixed: all 5 plans | EU option: AI Model Serving |
| Training on inputs | opt-out available: all 5 plans | no: AI Model Serving |
| DPA | not documented: Free, Plus, Premium / Super / Ultra, Teamsyes: Enterprise | yes: AI Model Serving |
| As of | 18 Oct 2026 | 18 Oct 2026 |
Plans compared
Comparing: Cursor Teams vs. STACKIT AI AI Model Serving
| Feature | CursorTeams | STACKIT AIAI Model Serving (usage-based) |
|---|---|---|
| differs by plan | not publicly documented | yes |
| Details: DPA available | not publicly documented (the public DPA refers to a Master Services Agreement with Anysphere; according to the Terms of Service an MSA applies only if the organisation has concluded one; the Privacy Policy mentions a processor role only generally 'For commercial uses of the Service') | yes (data processing agreement STACKIT Cloud Services, version 2.4.2, under Art. 28 GDPR; applies to all STACKIT Cloud Services booked under the terms of use, no product-specific annex for AI Model Serving) Provider’s DPA |
| differs by plan | opt-out available | no |
| Details: Training on inputs | Data Use (as of 2026-09-03): with Privacy Mode enabled, Customer Data is not used by Cursor for training; Cursor has zero data retention agreements with all model providers, and the providers do not store or train on the data. Exception: risk classifiers; where abuse detection is triggered, data may be stored for investigation. Non-ZDR models are labelled or require admin approval. Without Privacy Mode, Cursor may store codebase data, prompts, editor actions and code snippets and use them to train its own models. Teams: team-wide privacy settings; new members inherit the team setting; admins can enforce Privacy Mode. | Product page: 'your data and queries are neither stored nor used to train models' (translated). Documentation FAQ: 'We do not train any models using your data.' Service specification: apart from billing-relevant data, 'no customer data whatsoever is collected or analysed' (translated). DPA section 2.3: usage data (not content) is processed, among other things, to optimise and further develop the services and is anonymised; STACKIT may use anonymised usage data for its own purposes. |
| differs by plan | mixed | EU option |
| Details: Data residency | Privacy Policy (as of 2026-10-11): processing on servers in various jurisdictions, including the US; data of EEA users may be transferred to the US. Sub-processor list (Trust Center): 'Our infrastructure is primarily hosted on AWS US region with some latency critical services in Europe and Singapore.' Docs: data residency only for Enterprise; US-only (inference, processing, storage incl. backups in the US, 10% uplift); EU/UK inference only on request, further EU coverage in development. | Models run in region EU01 (price table and model documentation, API endpoint '...model-serving.eu01.onstackit.cloud'). Service description section 1.2: operation in STACKIT data centres in Germany and Austria (possibly further EU states in future). DPA section 2.5: processing within the EU/EEA; outside only where expressly agreed and Art. 44–49 GDPR are met. |
| list available | list available | |
| Details: List published |
|
|
| differs by plan | With Privacy Mode: ZDR with model… | Service specification |
| Details: Data retention | With Privacy Mode: ZDR with model providers; file contents are encrypted to reduce latency and cached only for the duration of the request. Cloud Agents: according to the documentation, repository contents are stored only during the agent run and deleted afterwards. Without Privacy Mode: inference providers may temporarily store inputs/outputs, which are deleted afterwards. Personal data in general: only as long as necessary (Privacy Policy); specific retention periods not publicly documented. | Service specification (annex on exportability): customer data is not stored permanently, no backup of requests; e-mail and subject ID are stored temporarily (30 days) in the service logs; changes to authentication tokens are recorded in the STACKIT audit log (retention period not quantified there). Documentation FAQ: 'We do not store any customer data from the requests.' DPA section 10.1: deletion of customer data after the end of the contract, deletion record on request. |
| differs by plan | from 30 USD/user/mo | per token |
| Details: Price | from USD 30 per user per month (pricing page: "$30+/user/month", excl. taxes); seat types can be combined. | Usage-based per 1 million tokens, region EU01 (product page, as of 2026-10-18; net prices according to terms of use section 11.3), input/output: gpt-oss 20b EUR 0.15/0.25; Gemma 4 31b, gpt-oss 120b, Llama 3.3 70B Instruct FP8, Qwen3-VL 235B (Plus) and Qwen3.8 27B EUR 0.45/0.65 each; Qwen3-VL 235B A22B (Premium) EUR 1.50/1.75. Embeddings (input only): e5 mistral 7b EUR 0.02; Qwen3 VL Embedding 8B EUR 0.08. Free trial access for 30 days after manual review of the registration (no permanently free plan). |
| yes | yes | |
| Details: AI features included | yes (as Premium, plus team marketplace, centralised billing, team-wide privacy settings, custom chat data retention, SAML/OIDC SSO; usage 4x to 50x Plus depending on seat tier) | yes (chat and embedding models as shared instances via an OpenAI-compatible REST API with API key; including Qwen3-VL 235B, Qwen3.8 27B, Llama 3.3 70B, gpt-oss 120b/20b, Gemma 4 31b; tool calling, image input for some models; open-source RAG template) |
| 6 sources | 9 sources | |
| Details: Sources with retrieval date |
|
|
Contract / DPA
- differs by plan
- Cursor Teamsnot publicly documented
- STACKIT AI AI Model Servingyes
- Cursor Teams
not publicly documented (the public DPA refers to a Master Services Agreement with Anysphere; according to the Terms of Service an MSA applies only if the organisation has concluded one; the Privacy Policy mentions a processor role only generally 'For commercial uses of the Service')
STACKIT AI AI Model Servingyes (data processing agreement STACKIT Cloud Services, version 2.4.2, under Art. 28 GDPR; applies to all STACKIT Cloud Services booked under the terms of use, no product-specific annex for AI Model Serving)
Provider’s DPA
Training
- differs by plan
- Cursor Teamsopt-out available
- STACKIT AI AI Model Servingno
- Cursor Teams
Data Use (as of 2026-09-03): with Privacy Mode enabled, Customer Data is not used by Cursor for training; Cursor has zero data retention agreements with all model providers, and the providers do not store or train on the data. Exception: risk classifiers; where abuse detection is triggered, data may be stored for investigation. Non-ZDR models are labelled or require admin approval. Without Privacy Mode, Cursor may store codebase data, prompts, editor actions and code snippets and use them to train its own models. Teams: team-wide privacy settings; new members inherit the team setting; admins can enforce Privacy Mode.
STACKIT AI AI Model ServingProduct page: 'your data and queries are neither stored nor used to train models' (translated). Documentation FAQ: 'We do not train any models using your data.' Service specification: apart from billing-relevant data, 'no customer data whatsoever is collected or analysed' (translated). DPA section 2.3: usage data (not content) is processed, among other things, to optimise and further develop the services and is anonymised; STACKIT may use anonymised usage data for its own purposes.
Hosting
- differs by plan
- Cursor Teamsmixed
- STACKIT AI AI Model ServingEU option
- Cursor Teams
Privacy Policy (as of 2026-10-11): processing on servers in various jurisdictions, including the US; data of EEA users may be transferred to the US. Sub-processor list (Trust Center): 'Our infrastructure is primarily hosted on AWS US region with some latency critical services in Europe and Singapore.' Docs: data residency only for Enterprise; US-only (inference, processing, storage incl. backups in the US, 10% uplift); EU/UK inference only on request, further EU coverage in development.
STACKIT AI AI Model ServingModels run in region EU01 (price table and model documentation, API endpoint '...model-serving.eu01.onstackit.cloud'). Service description section 1.2: operation in STACKIT data centres in Germany and Austria (possibly further EU states in future). DPA section 2.5: processing within the EU/EEA; outside only where expressly agreed and Art. 44–49 GDPR are met.
Sub-processors
- Cursor Teamslist available
- STACKIT AI AI Model Servinglist available
- Cursor Teams
- Amazon Web Services (primary cloud hosting; US, EU, Singapore)
- Further infrastructure: Microsoft Azure (secondary, US), Google Cloud Platform (US), Vercel (web hosting, US), Cloudflare (CDN/network security, worldwide)
- Model providers: OpenAI, Anthropic, Google Gemini (Vertex API), Meta (each worldwide), SpaceXAI (US), TypeSafe AI (US)
- Inference infrastructure: Fireworks (US, Canada, EU, UK), Baseten (US, Canada), Together (own models, with a ZDR agreement according to the list; US, Canada, UK), Nebius (UK), CoreWeave (US), Databricks (analytics database and inference, worldwide)
- Exa (web search, US), Datadog (logging/monitoring, US), WorkOS (SSO/SCIM, US)
STACKIT AI AI Model Serving- DPA Annex 2 (version 2.4.2), development, operations and consulting support for STACKIT Cloud Services as a whole: Anynines GmbH (DE), CGI Deutschland B.V. & Co. KG (DE), Cloud&Heat Technologies GmbH (DE), Exivity BV (NL), Freiheit.com Technologies GmbH (DE), Inovex GmbH (DE), ISD Feniqs GmbH (DE), Nextevolution GmbH (DE), Rewion GmbH (DE), Schwarz Digits Bulgaria EOOD (BG), Schwarz Digits IT KG (DE), Schwarz Digits Romania S.C.S. (RO), Schwarz IT S.L.U. (ES), SOCIAL BIGDATA COMUNICACION SL (ES)
- Which of these processors are actually used for AI Model Serving is not publicly documented; according to the documentation the model developers are only the source of the models (Hugging Face) and are not named as processors
Retention
- differs by plan
- Cursor TeamsWith Privacy Mode: ZDR with model…
- STACKIT AI AI Model ServingService specification
- Cursor TeamsWith Privacy Mode: ZDR with model providers; file contents are encrypted to reduce latency and cached only for the duration of the request. Cloud Agents: according to the documentation, repository contents are stored only during the agent run and deleted afterwards. Without Privacy Mode: inference providers may temporarily store inputs/outputs, which are deleted afterwards. Personal data in general: only as long as necessary (Privacy Policy); specific retention periods not publicly documented.STACKIT AI AI Model ServingService specification (annex on exportability): customer data is not stored permanently, no backup of requests; e-mail and subject ID are stored temporarily (30 days) in the service logs; changes to authentication tokens are recorded in the STACKIT audit log (retention period not quantified there). Documentation FAQ: 'We do not store any customer data from the requests.' DPA section 10.1: deletion of customer data after the end of the contract, deletion record on request.
Price
- differs by plan
- Cursor Teamsfrom 30 USD/user/mo
- STACKIT AI AI Model Servingper token
- Cursor Teams
from USD 30 per user per month (pricing page: "$30+/user/month", excl. taxes); seat types can be combined.
STACKIT AI AI Model ServingUsage-based per 1 million tokens, region EU01 (product page, as of 2026-10-18; net prices according to terms of use section 11.3), input/output: gpt-oss 20b EUR 0.15/0.25; Gemma 4 31b, gpt-oss 120b, Llama 3.3 70B Instruct FP8, Qwen3-VL 235B (Plus) and Qwen3.8 27B EUR 0.45/0.65 each; Qwen3-VL 235B A22B (Premium) EUR 1.50/1.75. Embeddings (input only): e5 mistral 7b EUR 0.02; Qwen3 VL Embedding 8B EUR 0.08. Free trial access for 30 days after manual review of the registration (no permanently free plan).
- Cursor Teamsyes
- STACKIT AI AI Model Servingyes
- Cursor Teams
yes (as Premium, plus team marketplace, centralised billing, team-wide privacy settings, custom chat data retention, SAML/OIDC SSO; usage 4x to 50x Plus depending on seat tier)
STACKIT AI AI Model Servingyes (chat and embedding models as shared instances via an OpenAI-compatible REST API with API key; including Qwen3-VL 235B, Qwen3.8 27B, Llama 3.3 70B, gpt-oss 120b/20b, Gemma 4 31b; tool calling, image input for some models; open-source RAG template)
Evidence
- Cursor Teams6 sources
- STACKIT AI AI Model Serving9 sources
- Cursor Teams
- PricingPricing page, retrieved 18 Oct 2026
- Security (as of 2026-08-25)Trust Center, retrieved 18 Oct 2026
- Data Use & Privacy Overview (as of 2026-09-03)Privacy notice, retrieved 18 Oct 2026
- Privacy Policy (as of 2026-10-11)Privacy notice, retrieved 18 Oct 2026
- Docs: Privacy and data governanceDocumentation, retrieved 18 Oct 2026
- Trust Center: Subprocessors (Vanta, 20 entries)Sub-processor list, retrieved 18 Oct 2026
STACKIT AI AI Model Serving- STACKIT AI Model Serving (product page with price table)Product page, retrieved 18 Oct 2026
- Service specification STACKIT AI Model Serving, version 1.3, valid from 18 September 2025Service description, retrieved 18 Oct 2026
- Service specification STACKIT AI Model Serving, version 1.4, valid from 6 November 2026Service description, retrieved 18 Oct 2026
- Data processing agreement STACKIT Cloud Services, version 2.4.2DPA, retrieved 18 Oct 2026
- Terms of use STACKIT Cloud, version 1.3.2, valid from 4 May 2026Terms and conditions, retrieved 18 Oct 2026
- Service description STACKIT Cloud, version 1.4, valid from 4 May 2026Service description, retrieved 18 Oct 2026
- AI Model Serving – FAQDocumentation, retrieved 18 Oct 2026
- AI Model Serving – Service descriptionDocumentation, retrieved 18 Oct 2026
- Available Shared Models of STACKIT AI Model ServingDocumentation, retrieved 18 Oct 2026
| Feature | Cursor | STACKIT AI |
|---|---|---|
| differs by tool |
| yesonly plan |
| Details: DPA available |
|
|
| differs by tool | opt-out availableall 5 plans | noonly plan |
| Details: Training on inputs |
|
|
| differs by tool | mixedall 5 plans | EU optiononly plan |
| Details: Data residency |
|
|
| list availableall 5 plans | list availableonly plan | |
| Details: List published |
|
|
| differs by tool | With Privacy Mode: ZDR with model…all 5 plans | Service specificationonly plan |
| Details: Data retention |
|
|
| differs by tool |
| per tokenonly plan |
| Details: Price |
|
|
| yesall 5 plans | yesonly plan | |
| Details: AI features included |
|
|
| 8 sources | 9 sources | |
| Details: Sources with retrieval date |
|
|
Contract / DPA
- differs by tool
- Cursor
- not publicly documentedFree (free of charge), Plus, Premium / Super / Ultra (Individual), Teams
- yesEnterprise
- STACKIT AIyesonly plan
- Cursor
- Free (free of charge)
not publicly documented (the public DPA refers to a Master Services Agreement with Anysphere; according to the Terms of Service an MSA applies only if the organisation has concluded one; the Privacy Policy mentions a processor role only generally 'For commercial uses of the Service')
- Plus
not publicly documented (the public DPA refers to a Master Services Agreement with Anysphere; according to the Terms of Service an MSA applies only if the organisation has concluded one; the Privacy Policy mentions a processor role only generally 'For commercial uses of the Service')
- Premium / Super / Ultra (Individual)
not publicly documented (the public DPA refers to a Master Services Agreement with Anysphere; according to the Terms of Service an MSA applies only if the organisation has concluded one; the Privacy Policy mentions a processor role only generally 'For commercial uses of the Service')
- Teams
not publicly documented (the public DPA refers to a Master Services Agreement with Anysphere; according to the Terms of Service an MSA applies only if the organisation has concluded one; the Privacy Policy mentions a processor role only generally 'For commercial uses of the Service')
- Enterprise
yes (public DPA, applies together with the Master Services Agreement; EU SCCs Module Two and UK Addendum)
Provider’s DPA
STACKIT AI- AI Model Serving (usage-based)
yes (data processing agreement STACKIT Cloud Services, version 2.4.2, under Art. 28 GDPR; applies to all STACKIT Cloud Services booked under the terms of use, no product-specific annex for AI Model Serving)
Provider’s DPA
- Free (free of charge)
Training
- differs by tool
- Cursoropt-out availableall 5 plans
- STACKIT AInoonly plan
- Cursor
- Free (free of charge)
Data Use (as of 2026-09-03): with Privacy Mode enabled, Customer Data is not used by Cursor for training; Cursor has zero data retention agreements with all model providers, and the providers do not store or train on the data. Exception: risk classifiers; where abuse detection is triggered, data may be stored for investigation. Non-ZDR models are labelled or require admin approval. Without Privacy Mode, Cursor may store codebase data, prompts, editor actions and code snippets and use them to train its own models. According to the Security page, Privacy Mode is also available for Free and can be enabled in the settings; default setting for individual accounts not publicly documented.
- Plus
Data Use (as of 2026-09-03): with Privacy Mode enabled, Customer Data is not used by Cursor for training; Cursor has zero data retention agreements with all model providers, and the providers do not store or train on the data. Exception: risk classifiers; where abuse detection is triggered, data may be stored for investigation. Non-ZDR models are labelled or require admin approval. Without Privacy Mode, Cursor may store codebase data, prompts, editor actions and code snippets and use them to train its own models. According to the Security page, Privacy Mode is available for Free and Pro and can be enabled in the settings; default setting for individual accounts not publicly documented.
- Premium / Super / Ultra (Individual)
Data Use (as of 2026-09-03): with Privacy Mode enabled, Customer Data is not used by Cursor for training; Cursor has zero data retention agreements with all model providers, and the providers do not store or train on the data. Exception: risk classifiers; where abuse detection is triggered, data may be stored for investigation. Non-ZDR models are labelled or require admin approval. Without Privacy Mode, Cursor may store codebase data, prompts, editor actions and code snippets and use them to train its own models. According to the Security page, Privacy Mode is available for Free and Pro and can be enabled in the settings; default setting for individual accounts not publicly documented.
- Teams
Data Use (as of 2026-09-03): with Privacy Mode enabled, Customer Data is not used by Cursor for training; Cursor has zero data retention agreements with all model providers, and the providers do not store or train on the data. Exception: risk classifiers; where abuse detection is triggered, data may be stored for investigation. Non-ZDR models are labelled or require admin approval. Without Privacy Mode, Cursor may store codebase data, prompts, editor actions and code snippets and use them to train its own models. Teams: team-wide privacy settings; new members inherit the team setting; admins can enforce Privacy Mode.
- Enterprise
Data Use (as of 2026-09-03): with Privacy Mode enabled, Customer Data is not used by Cursor for training; Cursor has zero data retention agreements with all model providers, and the providers do not store or train on the data. Exception: risk classifiers; where abuse detection is triggered, data may be stored for investigation. Non-ZDR models are labelled or require admin approval. Without Privacy Mode, Cursor may store codebase data, prompts, editor actions and code snippets and use them to train its own models. Enterprise: according to the documentation, Privacy Mode is enabled by default and enforceable; requests to models with provider-side storage (according to the documentation, Claude Fable 5 and 5.1) fail until an admin approves the retention policy.
STACKIT AI- AI Model Serving (usage-based)
Product page: 'your data and queries are neither stored nor used to train models' (translated). Documentation FAQ: 'We do not train any models using your data.' Service specification: apart from billing-relevant data, 'no customer data whatsoever is collected or analysed' (translated). DPA section 2.3: usage data (not content) is processed, among other things, to optimise and further develop the services and is anonymised; STACKIT may use anonymised usage data for its own purposes.
- Free (free of charge)
Hosting
- differs by tool
- Cursormixedall 5 plans
- STACKIT AIEU optiononly plan
- Cursor
- Free (free of charge)
Privacy Policy (as of 2026-10-11): processing on servers in various jurisdictions, including the US; data of EEA users may be transferred to the US. Sub-processor list (Trust Center): 'Our infrastructure is primarily hosted on AWS US region with some latency critical services in Europe and Singapore.' Docs: data residency only for Enterprise; US-only (inference, processing, storage incl. backups in the US, 10% uplift); EU/UK inference only on request, further EU coverage in development.
- Plus
Privacy Policy (as of 2026-10-11): processing on servers in various jurisdictions, including the US; data of EEA users may be transferred to the US. Sub-processor list (Trust Center): 'Our infrastructure is primarily hosted on AWS US region with some latency critical services in Europe and Singapore.' Docs: data residency only for Enterprise; US-only (inference, processing, storage incl. backups in the US, 10% uplift); EU/UK inference only on request, further EU coverage in development.
- Premium / Super / Ultra (Individual)
Privacy Policy (as of 2026-10-11): processing on servers in various jurisdictions, including the US; data of EEA users may be transferred to the US. Sub-processor list (Trust Center): 'Our infrastructure is primarily hosted on AWS US region with some latency critical services in Europe and Singapore.' Docs: data residency only for Enterprise; US-only (inference, processing, storage incl. backups in the US, 10% uplift); EU/UK inference only on request, further EU coverage in development.
- Teams
Privacy Policy (as of 2026-10-11): processing on servers in various jurisdictions, including the US; data of EEA users may be transferred to the US. Sub-processor list (Trust Center): 'Our infrastructure is primarily hosted on AWS US region with some latency critical services in Europe and Singapore.' Docs: data residency only for Enterprise; US-only (inference, processing, storage incl. backups in the US, 10% uplift); EU/UK inference only on request, further EU coverage in development.
- Enterprise
Privacy Policy (as of 2026-10-11): processing on servers in various jurisdictions, including the US; data of EEA users may be transferred to the US. Sub-processor list (Trust Center): 'Our infrastructure is primarily hosted on AWS US region with some latency critical services in Europe and Singapore.' Docs: data residency only for Enterprise; US-only (inference, processing, storage incl. backups in the US, 10% uplift); EU/UK inference only on request, further EU coverage in development.
STACKIT AI- AI Model Serving (usage-based)
Models run in region EU01 (price table and model documentation, API endpoint '...model-serving.eu01.onstackit.cloud'). Service description section 1.2: operation in STACKIT data centres in Germany and Austria (possibly further EU states in future). DPA section 2.5: processing within the EU/EEA; outside only where expressly agreed and Art. 44–49 GDPR are met.
- Free (free of charge)
Sub-processors
- Cursorlist availableall 5 plans
- STACKIT AIlist availableonly plan
- Cursor
- Free (free of charge)
- Amazon Web Services (primary cloud hosting; US, EU, Singapore)
- Further infrastructure: Microsoft Azure (secondary, US), Google Cloud Platform (US), Vercel (web hosting, US), Cloudflare (CDN/network security, worldwide)
- Model providers: OpenAI, Anthropic, Google Gemini (Vertex API), Meta (each worldwide), SpaceXAI (US), TypeSafe AI (US)
- Inference infrastructure: Fireworks (US, Canada, EU, UK), Baseten (US, Canada), Together (own models, with a ZDR agreement according to the list; US, Canada, UK), Nebius (UK), CoreWeave (US), Databricks (analytics database and inference, worldwide)
- Exa (web search, US), Datadog (logging/monitoring, US), WorkOS (SSO/SCIM, US)
- Plus
- Amazon Web Services (primary cloud hosting; US, EU, Singapore)
- Further infrastructure: Microsoft Azure (secondary, US), Google Cloud Platform (US), Vercel (web hosting, US), Cloudflare (CDN/network security, worldwide)
- Model providers: OpenAI, Anthropic, Google Gemini (Vertex API), Meta (each worldwide), SpaceXAI (US), TypeSafe AI (US)
- Inference infrastructure: Fireworks (US, Canada, EU, UK), Baseten (US, Canada), Together (own models, with a ZDR agreement according to the list; US, Canada, UK), Nebius (UK), CoreWeave (US), Databricks (analytics database and inference, worldwide)
- Exa (web search, US), Datadog (logging/monitoring, US), WorkOS (SSO/SCIM, US)
- Premium / Super / Ultra (Individual)
- Amazon Web Services (primary cloud hosting; US, EU, Singapore)
- Further infrastructure: Microsoft Azure (secondary, US), Google Cloud Platform (US), Vercel (web hosting, US), Cloudflare (CDN/network security, worldwide)
- Model providers: OpenAI, Anthropic, Google Gemini (Vertex API), Meta (each worldwide), SpaceXAI (US), TypeSafe AI (US)
- Inference infrastructure: Fireworks (US, Canada, EU, UK), Baseten (US, Canada), Together (own models, with a ZDR agreement according to the list; US, Canada, UK), Nebius (UK), CoreWeave (US), Databricks (analytics database and inference, worldwide)
- Exa (web search, US), Datadog (logging/monitoring, US), WorkOS (SSO/SCIM, US)
- Teams
- Amazon Web Services (primary cloud hosting; US, EU, Singapore)
- Further infrastructure: Microsoft Azure (secondary, US), Google Cloud Platform (US), Vercel (web hosting, US), Cloudflare (CDN/network security, worldwide)
- Model providers: OpenAI, Anthropic, Google Gemini (Vertex API), Meta (each worldwide), SpaceXAI (US), TypeSafe AI (US)
- Inference infrastructure: Fireworks (US, Canada, EU, UK), Baseten (US, Canada), Together (own models, with a ZDR agreement according to the list; US, Canada, UK), Nebius (UK), CoreWeave (US), Databricks (analytics database and inference, worldwide)
- Exa (web search, US), Datadog (logging/monitoring, US), WorkOS (SSO/SCIM, US)
- Enterprise
- Amazon Web Services (primary cloud hosting; US, EU, Singapore)
- Further infrastructure: Microsoft Azure (secondary, US), Google Cloud Platform (US), Vercel (web hosting, US), Cloudflare (CDN/network security, worldwide)
- Model providers: OpenAI, Anthropic, Google Gemini (Vertex API), Meta (each worldwide), SpaceXAI (US), TypeSafe AI (US)
- Inference infrastructure: Fireworks (US, Canada, EU, UK), Baseten (US, Canada), Together (own models, with a ZDR agreement according to the list; US, Canada, UK), Nebius (UK), CoreWeave (US), Databricks (analytics database and inference, worldwide)
- Exa (web search, US), Datadog (logging/monitoring, US), WorkOS (SSO/SCIM, US)
STACKIT AI- AI Model Serving (usage-based)
- DPA Annex 2 (version 2.4.2), development, operations and consulting support for STACKIT Cloud Services as a whole: Anynines GmbH (DE), CGI Deutschland B.V. & Co. KG (DE), Cloud&Heat Technologies GmbH (DE), Exivity BV (NL), Freiheit.com Technologies GmbH (DE), Inovex GmbH (DE), ISD Feniqs GmbH (DE), Nextevolution GmbH (DE), Rewion GmbH (DE), Schwarz Digits Bulgaria EOOD (BG), Schwarz Digits IT KG (DE), Schwarz Digits Romania S.C.S. (RO), Schwarz IT S.L.U. (ES), SOCIAL BIGDATA COMUNICACION SL (ES)
- Which of these processors are actually used for AI Model Serving is not publicly documented; according to the documentation the model developers are only the source of the models (Hugging Face) and are not named as processors
- Free (free of charge)
Retention
- differs by tool
- CursorWith Privacy Mode: ZDR with model…all 5 plans
- STACKIT AIService specificationonly plan
- Cursor
- Free (free of charge)With Privacy Mode: ZDR with model providers; file contents are encrypted to reduce latency and cached only for the duration of the request. Cloud Agents: according to the documentation, repository contents are stored only during the agent run and deleted afterwards. Without Privacy Mode: inference providers may temporarily store inputs/outputs, which are deleted afterwards. Personal data in general: only as long as necessary (Privacy Policy); specific retention periods not publicly documented.
- PlusWith Privacy Mode: ZDR with model providers; file contents are encrypted to reduce latency and cached only for the duration of the request. Cloud Agents: according to the documentation, repository contents are stored only during the agent run and deleted afterwards. Without Privacy Mode: inference providers may temporarily store inputs/outputs, which are deleted afterwards. Personal data in general: only as long as necessary (Privacy Policy); specific retention periods not publicly documented.
- Premium / Super / Ultra (Individual)With Privacy Mode: ZDR with model providers; file contents are encrypted to reduce latency and cached only for the duration of the request. Cloud Agents: according to the documentation, repository contents are stored only during the agent run and deleted afterwards. Without Privacy Mode: inference providers may temporarily store inputs/outputs, which are deleted afterwards. Personal data in general: only as long as necessary (Privacy Policy); specific retention periods not publicly documented.
- TeamsWith Privacy Mode: ZDR with model providers; file contents are encrypted to reduce latency and cached only for the duration of the request. Cloud Agents: according to the documentation, repository contents are stored only during the agent run and deleted afterwards. Without Privacy Mode: inference providers may temporarily store inputs/outputs, which are deleted afterwards. Personal data in general: only as long as necessary (Privacy Policy); specific retention periods not publicly documented.
- EnterpriseWith Privacy Mode: ZDR with model providers; file contents are encrypted to reduce latency and cached only for the duration of the request. Cloud Agents: according to the documentation, repository contents are stored only during the agent run and deleted afterwards. Without Privacy Mode: inference providers may temporarily store inputs/outputs, which are deleted afterwards. Personal data in general: only as long as necessary (Privacy Policy); specific retention periods not publicly documented.
STACKIT AI- AI Model Serving (usage-based)Service specification (annex on exportability): customer data is not stored permanently, no backup of requests; e-mail and subject ID are stored temporarily (30 days) in the service logs; changes to authentication tokens are recorded in the STACKIT audit log (retention period not quantified there). Documentation FAQ: 'We do not store any customer data from the requests.' DPA section 10.1: deletion of customer data after the end of the contract, deletion record on request.
Price
- differs by tool
- Cursor
- 0 USD/moFree (free of charge)
- 8 USD/moPlus
- several tiersPremium / Super / Ultra (Individual)
- from 30 USD/user/moTeams
- customEnterprise
- STACKIT AIper tokenonly plan
- Cursor
- Free (free of charge)
USD 0 per month (pricing page; prices excl. taxes such as VAT)
- Plus
USD 8 per month (pricing page, excl. taxes). According to the pricing page FAQ, subscriptions are sold via X Subscription Corp.
- Premium / Super / Ultra (Individual)
Premium USD 30, Super USD 100, Ultra USD 200 per month (pricing page, excl. taxes); third-party models are billed separately based on usage. According to the FAQ, sold via X Subscription Corp.
- Teams
from USD 30 per user per month (pricing page: "$30+/user/month", excl. taxes); seat types can be combined.
- Enterprise
Custom (pricing page: 'Custom'); invoice/PO billing available
STACKIT AI- AI Model Serving (usage-based)
Usage-based per 1 million tokens, region EU01 (product page, as of 2026-10-18; net prices according to terms of use section 11.3), input/output: gpt-oss 20b EUR 0.15/0.25; Gemma 4 31b, gpt-oss 120b, Llama 3.3 70B Instruct FP8, Qwen3-VL 235B (Plus) and Qwen3.8 27B EUR 0.45/0.65 each; Qwen3-VL 235B A22B (Premium) EUR 1.50/1.75. Embeddings (input only): e5 mistral 7b EUR 0.02; Qwen3 VL Embedding 8B EUR 0.08. Free trial access for 30 days after manual review of the registration (no permanently free plan).
- Free (free of charge)
- Cursoryesall 5 plans
- STACKIT AIyesonly plan
- Cursor
- Free (free of charge)
yes (limited AI usage, limited Agent and chat requests; Grok Chat/Imagine/Build limited)
- Plus
yes (shared AI usage pool; according to the pricing page, Grok models only; Grok Chat, Voice, Imagine, Build, Grok API access; plus X features)
- Premium / Super / Ultra (Individual)
yes (Premium 4x, Super 15x, Ultra 50x usage compared with Plus; Cloud Agents and automations, Bugbot, third-party models at API list price, Grok Bot, video generation; plus X features)
- Teams
yes (as Premium, plus team marketplace, centralised billing, team-wide privacy settings, custom chat data retention, SAML/OIDC SSO; usage 4x to 50x Plus depending on seat tier)
- Enterprise
yes (as Teams, plus pooled usage, invoice/PO, SCIM, access, network and model controls, audit logs, Priority Support; optional US data residency)
STACKIT AI- AI Model Serving (usage-based)
yes (chat and embedding models as shared instances via an OpenAI-compatible REST API with API key; including Qwen3-VL 235B, Qwen3.8 27B, Llama 3.3 70B, gpt-oss 120b/20b, Gemma 4 31b; tool calling, image input for some models; open-source RAG template)
- Free (free of charge)
Evidence
- Cursor8 sources
- STACKIT AI9 sources
- Cursor
- PricingPricing page, retrieved 18 Oct 2026
- Security (as of 2026-08-25)Trust Center, retrieved 18 Oct 2026
- Data Use & Privacy Overview (as of 2026-09-03)Privacy notice, retrieved 18 Oct 2026
- Privacy Policy (as of 2026-10-11)Privacy notice, retrieved 18 Oct 2026
- Data Processing Addendum (as of 2025-11-05)DPA, retrieved 18 Oct 2026
- Docs: Privacy and data governanceDocumentation, retrieved 18 Oct 2026
- Trust Center: Subprocessors (Vanta, 20 entries)Sub-processor list, retrieved 18 Oct 2026
- SpaceXAI Bundled Subscription Terms (X Subscription Corp)Terms and conditions, retrieved 18 Oct 2026
STACKIT AI- STACKIT AI Model Serving (product page with price table)Product page, retrieved 18 Oct 2026
- Service specification STACKIT AI Model Serving, version 1.3, valid from 18 September 2025Service description, retrieved 18 Oct 2026
- Service specification STACKIT AI Model Serving, version 1.4, valid from 6 November 2026Service description, retrieved 18 Oct 2026
- Data processing agreement STACKIT Cloud Services, version 2.4.2DPA, retrieved 18 Oct 2026
- Terms of use STACKIT Cloud, version 1.3.2, valid from 4 May 2026Terms and conditions, retrieved 18 Oct 2026
- Service description STACKIT Cloud, version 1.4, valid from 4 May 2026Service description, retrieved 18 Oct 2026
- AI Model Serving – FAQDocumentation, retrieved 18 Oct 2026
- AI Model Serving – Service descriptionDocumentation, retrieved 18 Oct 2026
- Available Shared Models of STACKIT AI Model ServingDocumentation, retrieved 18 Oct 2026
not documentedmeans: not publicly documented, not verifiable in any provider source. It does not mean “no”.
No guarantee of accuracy, not legal advice. Each entry is valid as of the date shown. Open points are listed on the detail pages.
More comparisons
Cursor compared with
Individual pages: Cursor, STACKIT AI. All tools in the category: Software development.