OpenAI API
KI-Anbieter
Hosted models for text and image inputs, with Responses API access and client SDKs. · Nachweis erneut prüfen
Preisbedingungen
Nutzungsbasierte API-Preise
Die Plattform wählen, die KI-Modelle bereitstellt; API-Zugang, Abrechnung und Betriebsbedingungen getrennt von einzelnen Modellspezifikationen vergleichen.
KI-Anbieter
Hosted models for text and image inputs, with Responses API access and client SDKs. · Nachweis erneut prüfen
Nutzungsbasierte API-Preise
KI-Anbieter
Programmatic access to Claude models through a REST API. · Nachweis erneut prüfen
Nutzungsbasierte API-Preise
KI-Anbieter
A developer API catalogue with text, multimodal, audio and image model families. · Nachweis erneut prüfen
Nutzungsbasierte API-Preise
KI-Anbieter
Mistral AI provides an API described in its client documentation for chat completion and embeddings.
Nutzungsbasierte API-Preise
KI-Anbieter
Groq offers a REST API accessed by an official Python client.
KI-Anbieter
Cohere documents a Python SDK for access to its models across multiple platforms.
KI-Anbieter
OpenRouter documents a Python SDK for accessing models across providers.
KI-Anbieter
The official xAI Python SDK describes access to xAI APIs.
Nutzungsbasierte API-Preise
KI-Anbieter
The Perplexity Python library describes access to the Perplexity REST API.
Nutzungsbasierte API-Preise
KI-Anbieter
The publisher Python library describes access to the Cerebras REST API.
KI-Anbieter
The publisher documents serverless text generation through Together AI.
Nutzungsbasierte API-Preise
KI-Anbieter
The official Fireworks Python client documents access to the Fireworks REST API and includes a chat-completion example.
KI-Anbieter
The Baseten Truss project describes a CLI for deploying and serving machine-learning models on Baseten.
KI-Anbieter
The official SambaNova Python client describes access to the SambaNova REST API with synchronous and asynchronous clients.
KI-Anbieter
NVIDIA describes NIM as an inference microservice used to deploy and serve large language models.
KI-Anbieter
The Replicate Python client describes running models from Python code or a notebook.
Nutzungsbasierte API-Preise
KI-Anbieter
The DeepInfra Python repository describes a client for inference and sandbox operations.
Nutzungsbasierte API-Preise
KI-Anbieter
The IBM watsonx.ai Node.js SDK describes programmatic interaction with the watsonx.ai service.
KI-Anbieter
The InferenceClient guide describes routing requests through Hugging Face to third-party inference providers with usage billed to a Hugging Face account.
KI-Anbieter
The Nebius cookbook describes building applications with open models on Token Factory.
Nutzungsbasierte API-Preise
| Anforderung | OpenAI API | Anthropic Claude API | Google Gemini API | Mistral AI API | Groq API | Cohere API | OpenRouter API | xAI API | Perplexity API Platform | Cerebras Inference | Together AI | Fireworks AI API | Baseten | SambaCloud | NVIDIA NIM | Replicate | DeepInfra | IBM watsonx.ai | Hugging Face Inference Providers | Nebius Token Factory |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Preise | Nutzungsbasierte API-Preise Preis erneut prüfen | Nutzungsbasierte API-Preise Preis erneut prüfen | Nutzungsbasierte API-Preise Preis erneut prüfen | Nutzungsbasierte API-Preise | Preis nicht geprüft | Preis nicht geprüft | Preis nicht geprüft | Nutzungsbasierte API-Preise | Nutzungsbasierte API-Preise | Preis nicht geprüft | Nutzungsbasierte API-Preise | Preis nicht geprüft | Preis nicht geprüft | Preis nicht geprüft | Preis nicht geprüft | Nutzungsbasierte API-Preise | Nutzungsbasierte API-Preise | Preis nicht geprüft | Preis nicht geprüft | Nutzungsbasierte API-Preise |
| Für wen | Hosted models for text and image inputs, with Responses API access and client SDKs. Geprüft am 18. Okt. 2026 · OpenAI API: official API documentation | Programmatic access to Claude models through a REST API. Geprüft am 18. Okt. 2026 · Anthropic Claude API: official API documentation | A developer API catalogue with text, multimodal, audio and image model families. Geprüft am 18. Okt. 2026 · Google Gemini API: official API documentation | Mistral AI provides an API described in its client documentation for chat completion and embeddings. Supplier-maintained SDK documentation captured 3 November 2026. SDK description, not a service price, model guarantee or benchmark. No API request or account was created. Geprüft am 3. Nov. 2026 · Mistral AI API: mistral-readme | Groq offers a REST API accessed by an official Python client. Supplier-maintained SDK documentation captured 3 November 2026. SDK description, not a service price, model guarantee or benchmark. No API request or account was created. Geprüft am 3. Nov. 2026 · Groq API: groq-readme | Cohere documents a Python SDK for access to its models across multiple platforms. Geprüft am 3. Nov. 2026 · Cohere API: cohere-readme | OpenRouter documents a Python SDK for accessing models across providers. Geprüft am 3. Nov. 2026 · OpenRouter API: openrouter-readme | The official xAI Python SDK describes access to xAI APIs. Geprüft am 3. Nov. 2026 · xAI API: xai-sdk | The Perplexity Python library describes access to the Perplexity REST API. Geprüft am 3. Nov. 2026 · Perplexity API Platform: perplexity-sdk | The publisher Python library describes access to the Cerebras REST API. Geprüft am 3. Nov. 2026 · Cerebras Inference: cerebras-docs | The publisher documents serverless text generation through Together AI. Geprüft am 3. Nov. 2026 · Together AI: together-docs | The official Fireworks Python client documents access to the Fireworks REST API and includes a chat-completion example. Geprüft am 4. Nov. 2026 · Fireworks AI API: fireworks-api-integration-source | The Baseten Truss project describes a CLI for deploying and serving machine-learning models on Baseten. Geprüft am 4. Nov. 2026 · Baseten: baseten-api-integration-source | The official SambaNova Python client describes access to the SambaNova REST API with synchronous and asynchronous clients. Geprüft am 4. Nov. 2026 · SambaCloud: sambacloud-integration-source | NVIDIA describes NIM as an inference microservice used to deploy and serve large language models. Geprüft am 4. Nov. 2026 · NVIDIA NIM: nim-blueprint-source | The Replicate Python client describes running models from Python code or a notebook. Geprüft am 4. Nov. 2026 · Replicate: replicate-api-integration-source | The DeepInfra Python repository describes a client for inference and sandbox operations. Geprüft am 4. Nov. 2026 · DeepInfra: deepinfra-api-integration-source | The IBM watsonx.ai Node.js SDK describes programmatic interaction with the watsonx.ai service. Geprüft am 4. Nov. 2026 · IBM watsonx.ai: ibm-watsonx-ai-source | The InferenceClient guide describes routing requests through Hugging Face to third-party inference providers with usage billed to a Hugging Face account. Geprüft am 4. Nov. 2026 · Hugging Face Inference Providers: huggingface-inference-providers-source | The Nebius cookbook describes building applications with open models on Token Factory. Geprüft am 4. Nov. 2026 · Nebius Token Factory: nebius-token-factory-source |
| Abdeckungsgrenzen | Capabilities, rates and tool support vary by model. A model catalogue entry does not establish access through a consumer subscription. Geprüft am 18. Okt. 2026 · OpenAI API: official API documentation | Direct API access requires a Console account and credentials; cloud-platform access has its own configuration. Geprüft am 18. Okt. 2026 · Anthropic Claude API: official API documentation | Model availability differs by family. The documentation restricts Gemini 2.5 access to prior users and directs new projects to newer models. Geprüft am 18. Okt. 2026 · Google Gemini API: official API documentation | The documented Python client requires a Mistral API key. A credential requirement only; this does not establish account eligibility, price, billing, quota or data retention. Geprüft am 3. Nov. 2026 · Mistral AI API: mistral-readme | The documented Python library supports Python 3.10 or later. This is a client runtime requirement, not a statement about all clients, model availability, rates, quota or hosted service terms. Geprüft am 3. Nov. 2026 · Groq API: groq-readme | The SDK documentation says Generate and Rerank models on OCI require fine-tuning and deployment to a dedicated endpoint. Geprüft am 3. Nov. 2026 · Cohere API: cohere-readme | The OpenRouter Python SDK requires Python 3.10 or higher. Geprüft am 3. Nov. 2026 · OpenRouter API: openrouter-readme | The xAI Python SDK describes synchronous and asynchronous clients. Geprüft am 3. Nov. 2026 · xAI API: xai-sdk | The Perplexity Python library describes synchronous and asynchronous clients. Geprüft am 3. Nov. 2026 · Perplexity API Platform: perplexity-sdk | The documented Python library offers synchronous and asynchronous clients. Geprüft am 3. Nov. 2026 · Cerebras Inference: cerebras-docs | The publisher describes Python SDK v2 examples for Together AI. Geprüft am 3. Nov. 2026 · Together AI: together-docs | The captured SDK documentation says training calls require a training-scoped Fireworks API key. Geprüft am 4. Nov. 2026 · Fireworks AI API: fireworks-api-integration-source | The Truss quickstart asks users to sign up or sign in to Baseten before deploying its example. Geprüft am 4. Nov. 2026 · Baseten: baseten-api-integration-source | The documented client example reads a SambaNova API key from an environment variable. Geprüft am 4. Nov. 2026 · SambaCloud: sambacloud-integration-source | The captured repository says its cloud service provider sample repository is being deprecated as the canonical home for those samples. Geprüft am 4. Nov. 2026 · NVIDIA NIM: nvidia-nim-integration-source | The captured client documentation says version 1.0.0 changed the default return type for models that output files. Geprüft am 4. Nov. 2026 · Replicate: replicate-api-integration-source | The captured client README describes authentication using a DeepInfra API key. Geprüft am 4. Nov. 2026 · DeepInfra: deepinfra-api-integration-source | The documented Node SDK requires access to watsonx.ai on IBM Cloud or Cloud Pak for Data and identifies Node.js 20 or newer as tested. Geprüft am 4. Nov. 2026 · IBM watsonx.ai: ibm-watsonx-ai-source | The same client guide describes a separate direct-access option using the customer’s own provider API key. Geprüft am 4. Nov. 2026 · Hugging Face Inference Providers: huggingface-inference-providers-source | The captured quickstart requires an API key and selecting a text model from the Token Factory model catalog. Publisher cookbook at captured repository revision, 4 November 2026. This is a specific quickstart and does not establish all available models, supported regions, service pricing or account entitlements. No account or model request. Geprüft am 4. Nov. 2026 · Nebius Token Factory: nebius-token-factory-source |
| API-Schnittstellen | Responses API and client SDKs Geprüft am 18. Okt. 2026 · OpenAI API: official API documentation | Messages API, Message Batches API, Token Counting API, Models API Geprüft am 18. Okt. 2026 · Anthropic Claude API: official API documentation | Gemini API Geprüft am 18. Okt. 2026 · Google Gemini API: official API documentation | The documented API includes chat completion and embeddings interfaces. Project SDK summary; exact models, regions and endpoint entitlements must be checked. No API request was made. Geprüft am 3. Nov. 2026 · Mistral AI API: mistral-readme | The documented Python client supports streaming responses using server-sent events. SDK documentation, not a performance guarantee or proof that every model or operation supports identical streaming. No API request was made. Geprüft am 3. Nov. 2026 · Groq API: groq-readme | The Python SDK documents streaming endpoints and chat streaming. Geprüft am 3. Nov. 2026 · Cohere API: cohere-readme | The SDK documentation gives an example using the Responses API. Geprüft am 3. Nov. 2026 · OpenRouter API: openrouter-readme | Responses API, Image and video generation, Files API, Batch API, Voice API, Tokenization, Model and account lookup Official TypeScript SDK README (xai-org/xai-sdk-ts, release v0.2.8, commit ef80b018, Apache-2.0) captured 9 November 2026; supplier documentation, not an API call, deployment or independent test. Lists what this SDK release covers, not every API the provider offers; the README marks the SDK experimental, so its interfaces may change before 1.0. Geprüft am 11. Nov. 2026 · xAI API: xai-api-project | Orchestrator API, Agent API, Search API, Embeddings API, Decisions API Perplexity API pricing page (docs.perplexity.ai) captured 9 November 2026; prices in USD; supplier statement, not a hands-on test. These are the APIs with a pricing section on the page, not a complete API reference; features and rates differ by API. Geprüft am 10. Nov. 2026 · Perplexity API Platform: perplexity-api-pricing-20261109b | Nicht geprüft | OpenAI-compatible serverless text generation Publisher skills README (togethercomputer/skills) captured 3 November 2026; supplier description of the chat-completions skill, not an API call or compatibility test. Which OpenAI endpoints and parameters are supported is not established. Geprüft am 11. Nov. 2026 · Together AI: together-docs | Nicht geprüft | Nicht geprüft | Nicht geprüft | A deployed NIM service exposes OpenAI-compatible REST endpoints for health check, completions, chat completions and model listing. Publisher blueprint README at the captured repository revision, 9 November 2026; it describes the container used in the blueprint, not every NIM. Reference code licence does not establish hosted-service, model, image or trademark terms. No account, deployment or request. Geprüft am 10. Nov. 2026 · NVIDIA NIM: nvidia-nim-blueprint-project | HTTP API, Node.js client library Replicate Node.js client README (replicate/replicate-javascript, release v1.4.0, commit f863dd2c, Apache-2.0) captured 9 November 2026; supplier documentation, not a test. The README says the library cannot call the API directly from a browser. Geprüft am 10. Nov. 2026 · Replicate: replicate-api-project | An OpenAI-compatible API at https://api.deepinfra.com/v1/openai; the documentation says an existing OpenAI SDK can be pointed at it without code changes. DeepInfra documentation start page captured 9 November 2026. Compatibility is the supplier's statement; individual models and features may differ and no request was made. Geprüft am 9. Nov. 2026 · DeepInfra: deepinfra-api-docs-20261109b | The documented Node.js SDK authenticates with an IAM API key, a bearer token, or IBM watsonx.ai software credentials set through environment variables. Publisher SDK documentation at captured repository revision, 9 November 2026. Client-specific authentication only; it does not establish model, service, plan or trademark terms. No account or model request. Geprüft am 10. Nov. 2026 · IBM watsonx.ai: ibm-watsonx-ai-project | Nicht geprüft | The captured cookbook shows a Python client example configured with the Token Factory API base URL. Geprüft am 4. Nov. 2026 · Nebius Token Factory: nebius-token-factory-source |
| Modellkatalog | GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna Geprüft am 18. Okt. 2026 · OpenAI API: official API documentation | Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5.5, Claude Haiku 5.5 Geprüft am 18. Okt. 2026 · Anthropic Claude API: official API documentation | Gemini 3.8 Flash, Gemini 3.5 Flash-Lite Geprüft am 18. Okt. 2026 · Google Gemini API: official API documentation | Mistral Large 4, OCR 4.1, Z.ai GLM 5.3, Z.ai GLM 5.3 Flash, DeepSeek V4.1 Flash 1017, Mistral Small 4 Featured models on the supplier Models Overview page captured 9 November 2026; the page lists further models, versions and licences. Z.ai GLM and DeepSeek are described as third-party models. Account and regional access were not verified. Geprüft am 10. Nov. 2026 · Mistral AI API: mistral-api-models-20261109b | Nicht geprüft | Nicht geprüft | Nicht geprüft | A separate catalogue for chat and image-understanding models lists modalities, aliases, token pricing and supported reasoning efforts. Official TypeScript SDK README (xai-org/xai-sdk-ts, release v0.2.8, commit ef80b018, Apache-2.0) captured 9 November 2026; supplier documentation, not an API call, deployment or independent test. Describes how the API and SDK expose model information; it does not name the currently available models. Check the provider’s model documentation for the live list. Geprüft am 11. Nov. 2026 · xAI API: xai-api-project | Agent API models come from third-party providers including OpenAI, Anthropic, Google, xAI, Z.AI, Moonshot AI and NVIDIA. Perplexity API pricing page (docs.perplexity.ai) captured 9 November 2026; prices in USD; supplier statement, not a hands-on test. Provider list only; the page links to a separate Agent API Models page for the individual models, which this capture does not include. The page also lists Perplexity-named embedding and decision models for other APIs. Geprüft am 10. Nov. 2026 · Perplexity API Platform: perplexity-api-pricing-20261109b | Nicht geprüft | Kimi K3, GLM-5.3, GLM-5.3-Flash Together AI pricing page captured 9 November 2026 (amounts shown with a $ sign; the page states no ISO currency); supplier statement, not a hands-on test. Three of the serverless chat models listed in the pricing table; the table lists more and changes often, so it is not the complete catalogue. Geprüft am 11. Nov. 2026 · Together AI: together-ai-pricing-20261109b | Nicht geprüft | Nicht geprüft | Nicht geprüft | Nicht geprüft | Thousands of open-source models contributed by the community, plus a wide variety of proprietary models. Replicate pricing page captured 9 November 2026 (amounts shown with a $ sign; the page states no ISO currency); supplier statement, not a hands-on test. Catalogue size is the supplier’s description and no model count is stated; individual models carry their own licences and terms. Geprüft am 10. Nov. 2026 · Replicate: replicate-api-pricing-20261109b | 100+ LLMs through an OpenAI-compatible API; Vision and OCR models; Embedding and reranker models; Image and video generation (FLUX, Stable Diffusion); Speech recognition (Whisper) and text-to-speech models DeepInfra documentation start page captured 9 November 2026. Categories of models as the page describes them, not a complete or current model list; availability and rates differ per model. Supplier statement, not a hands-on test. Geprüft am 9. Nov. 2026 · DeepInfra: deepinfra-api-docs-20261109b | Nicht geprüft | Nicht geprüft | MiniMax M3, Kimi K3, DeepSeek V4 Pro Models featured on the Token Factory product page captured 9 November 2026; the live model catalogue lists more models and was not captured. Availability, regions and prices per model need a separate check. Geprüft am 10. Nov. 2026 · Nebius Token Factory: nebius-token-factory-tf-20261109b |
| Abrechnungsgrundlage | Model-dependent input and output token rates; tools can carry additional charges. Example rates belong to GPT-6.1 Sol, not every model. See each exact model and its service tier. Geprüft am 18. Okt. 2026 · OpenAI API: official API documentation | Model-dependent input and output token charges, with separate cache and batch pricing. Geprüft am 18. Okt. 2026 · Anthropic Claude API: official API documentation | Model-specific free and paid tiers, with standard, batch and other service rates. Free and paid terms differ. Account eligibility, tool charges and dated rates need a model-specific check. Geprüft am 18. Okt. 2026 · Google Gemini API: official API documentation | Most models are billed per million input and output tokens, with batch and cached-input discounts; OCR, speech and tool APIs use other units. Supplier pricing-page FAQ captured 9 November 2026. The quoted example rate belongs to one model and is not repeated here; per-model rates, taxes and account eligibility need a model-specific check. Supplier statement, not a hands-on test; no account or API request. Geprüft am 10. Nov. 2026 · Mistral AI API: mistral-api-pricing-20261109b | Nicht geprüft | Nicht geprüft | Nicht geprüft | Usage is priced per token, with token prices reported in USD cents per 100 million tokens. Official TypeScript SDK README (xai-org/xai-sdk-ts, release v0.2.8, commit ef80b018, Apache-2.0) captured 9 November 2026; supplier documentation, not an API call, deployment or independent test. The same README says the Batch API processes requests asynchronously at a reduced price and that the "priority" or "fast" service tier costs more; it states no actual rates. Current rates and tool, image, video and voice charges must be checked in the provider’s pricing documentation. Geprüft am 11. Nov. 2026 · xAI API: xai-api-project | Orchestrator API usage is billed per token at each model’s published input and output rates, with no per-request fees. Perplexity API pricing page (docs.perplexity.ai) captured 9 November 2026; prices in USD; supplier statement, not a hands-on test. The same page prices the Agent API per token at each model’s published rates and lists separate charges: Search API $5.00 per 1,000 successful requests ($1.00 with Fast Search) and per-invocation Agent API tools (for example standard web search $0.0025). Per-model rates are on separate model pages that this capture does not include. Geprüft am 10. Nov. 2026 · Perplexity API Platform: perplexity-api-pricing-20261109b | Nicht geprüft | Models are priced per 1 million tokens, with separate input and output prices for each model. Together AI pricing page captured 9 November 2026 (amounts shown with a $ sign; the page states no ISO currency); supplier statement, not a hands-on test. Serverless Inference pricing table (Chat tab): rates vary by model, some models list a cached-input price, and the table header also lists a Batch API price. Image, video, audio and other modalities use other units. Dedicated endpoints, provisioned throughput and GPU clusters are priced separately and not captured. Geprüft am 11. Nov. 2026 · Together AI: together-ai-pricing-20261109b | Nicht geprüft | Nicht geprüft | Nicht geprüft | Nicht geprüft | Public models are mostly billed by run time at a per-second price that depends on the hardware used; some models are billed by input and output. Replicate pricing page captured 9 November 2026 (amounts shown with a $ sign; the page states no ISO currency); supplier statement, not a hands-on test. Private models on dedicated hardware are billed for all online time (setup, idle and active), except fast-booting fine-tunes. The page lists per-second and hourly hardware prices and per-output example prices; the live price of a given model is shown on its own page. Geprüft am 10. Nov. 2026 · Replicate: replicate-api-pricing-20261109b | Some language models are billed per token and most other models by inference execution time; the pricing page says there are no long-term contracts or upfront costs. DeepInfra pricing page captured 9 November 2026. Rates differ by model and are listed per model on the page; dedicated GPU deployments and embeddings have separate rate tables. The same page separately requires a card or pre-payment before use. Supplier statement, not a hands-on test; amounts are not restated here because they change. Geprüft am 9. Nov. 2026 · DeepInfra: deepinfra-api-pricing-20261109b | Nicht geprüft | Nicht geprüft | Serverless text generation is billed per input, cached input and output token; Dedicated Endpoints are billed per GPU hour. Nebius Token Factory documentation (Cached Tokens) captured 9 November 2026. Per-model token prices and Dedicated Endpoint GPU-hour rates are on pages that were not captured, so no amount is stated. Supplier statement, not a hands-on test; no account or API request. Geprüft am 10. Nov. 2026 · Nebius Token Factory: nebius-token-factory-cached-20261109b |
| Datenkontrollen | Nicht geprüft | Nicht geprüft | Nicht geprüft | Under the commercial terms Mistral AI does not train its models on Customer Data or Outputs except in listed cases: products with opt-in or opt-out training settings, feedback, Order Form provisions and Labs or Preview Models. Mistral Commercial Terms of Service captured 9 November 2026. Applies to commercial customers; consumer plans and individual products carry their own settings. Supplier terms, not legal advice; no account created. Geprüft am 10. Nov. 2026 · Mistral AI API: mistral-api-cterms-20261109b | Nicht geprüft | Nicht geprüft | Nicht geprüft | By default the SDK sends store: false, asking that responses not be stored, unless the caller opts in; this differs from the API’s own default. Official TypeScript SDK README (xai-org/xai-sdk-ts, release v0.2.8, commit ef80b018, Apache-2.0) captured 9 November 2026; supplier documentation, not an API call, deployment or independent test. Client behaviour of this SDK only. Retention, training use and other data terms of the service are governed by the provider’s terms and settings, which this capture does not cover. Geprüft am 11. Nov. 2026 · xAI API: xai-api-project | Customer content is not used to train generative AI models by default; zero data retention is not the default and is available only to eligible customers under a written agreement. Perplexity API Privacy & Security page (docs.perplexity.ai) captured 9 November 2026; supplier statement, not a hands-on test. The page says retention, use and security practices vary by API, model and account configuration, that third-party model providers may require limited retention, and that it does not modify the API Terms of Service, which control. The page also notes that certain APIs may be subject to product-specific terms. Geprüft am 10. Nov. 2026 · Perplexity API Platform: perplexity-api-privacy-20261109b | Nicht geprüft | By default prompts and responses are stored and may be used for product improvements; organization admins can turn storage off to enable zero data retention. Together AI Privacy and security documentation captured 9 November 2026; supplier statement, not an audit or test. The page also says training on customer data is opt-in and off by default, and that some passthrough models forward prompts to third-party providers who apply their own data policies. Geprüft am 11. Nov. 2026 · Together AI: together-ai-privacy-and-security-20261109b | Nicht geprüft | Nicht geprüft | Nicht geprüft | Nicht geprüft | The customer grants Replicate a licence to use, store and process its customer data to the extent necessary to provide the output, train and generate customer derivative models, provide the services and compile resultant data. Replicate Terms of service (last update April 1, 2026) section 5.2, captured 9 November 2026; supplier statement, not a hands-on test. Section 5.1 says the customer owns its Customer Data as between the parties; section 5.5 lets Replicate analyse resultant data to improve its services. Retention periods are not stated in the terms. Community Models shared on the platform carry further licences. Geprüft am 10. Nov. 2026 · Replicate: replicate-api-terms-20261109b | Input data is not stored to disk during inference and outputs are not stored; DeepInfra says it does not train on or share submitted data, with exceptions for Google and Anthropic models. DeepInfra Data Privacy documentation captured 9 November 2026. The page also says image-generation outputs are stored briefly and that bulk inference APIs may store data longer in encrypted form. Supplier statement, not an audit or certification. Geprüft am 9. Nov. 2026 · DeepInfra: deepinfra-api-data-20261109b | Nicht geprüft | Nicht geprüft | Public serverless endpoints are not intended for production use cases that depend on a stable region; for a guaranteed region, a Dedicated endpoint in the region of your choice always processes requests there. Nebius Token Factory documentation (Public Serverless Endpoints) captured 9 November 2026. Processing region only; retention, logging and training terms are separate. Supplier statement, not a hands-on test. Geprüft am 10. Nov. 2026 · Nebius Token Factory: nebius-token-factory-serverless-20261109b |
| Verfügbarkeit | Request and token rate limits depend on the account usage tier. Documentation check only; this catalogue does not verify access for a particular account. Geprüft am 18. Okt. 2026 · OpenAI API: official API documentation | Console credentials or configured Workload Identity Federation are prerequisites. Geprüft am 18. Okt. 2026 · Anthropic Claude API: official API documentation | Gemini 2.5 access is limited to prior users; current model status must be checked individually. Geprüft am 18. Okt. 2026 · Google Gemini API: official API documentation | API keys and API use require billing activation with pay-as-you-go; the Free mode covers Vibe and Studio but not the API. Supplier Admin documentation (Usage and limits) captured 9 November 2026. Plan terms and included usage can change; no account was created and no API request made. Geprüft am 10. Nov. 2026 · Mistral AI API: mistral-api-tier-20261109b | Nicht geprüft | Nicht geprüft | Nicht geprüft | A priority (or fast) service tier gives faster responses at a higher price. Official TypeScript SDK README (xai-org/xai-sdk-ts, release v0.2.8, commit ef80b018, Apache-2.0) captured 9 November 2026; supplier documentation, not an API call, deployment or independent test. Rates for the tier are on the provider’s Priority Processing page, which this capture does not include. Which models offer a fast deployment is not stated. Geprüft am 11. Nov. 2026 · xAI API: xai-api-project | API credits can be purchased through AWS Marketplace, and custom enterprise pricing is available through a sales enquiry. Perplexity API pricing page (docs.perplexity.ai) captured 9 November 2026; prices in USD; supplier statement, not a hands-on test. Purchase channels only; no custom or enterprise price is stated. Rate limits and usage tiers are on a separate page that this capture does not include. Geprüft am 10. Nov. 2026 · Perplexity API Platform: perplexity-api-pricing-20261109b | Nicht geprüft | Serverless performance is best-effort and requests may be limited under high demand; committed throughput with reliability guarantees is offered separately as provisioned throughput. Together AI serverless rate-limits documentation captured 9 November 2026; supplier statement, not a load test. Numeric limits are not stated on the page; the page says custom limits or access restrictions can apply to models in high demand. Geprüft am 11. Nov. 2026 · Together AI: together-ai-rate-limits-20261109b | Nicht geprüft | Nicht geprüft | Nicht geprüft | NVIDIA publishes reference deployments of NIM for Kubernetes and other production environments; the repository says the code is not validated on all platforms and carries no enterprise support. Publisher integration README at the captured repository revision, 9 November 2026. Sample and reference code only; official supported releases are distributed separately. No account, deployment or benchmark. Geprüft am 10. Nov. 2026 · NVIDIA NIM: nvidia-nim-project | Featured models can be tried for free; some features need billing to be set up. Replicate Terms of service (last update April 1, 2026) section 2.1(c), captured 9 November 2026; supplier statement, not a hands-on test. Which models are featured and which features need billing are not listed in the clause. The terms describe a Payment in Arrears and a Prepaid Credit billing model. Geprüft am 10. Nov. 2026 · Replicate: replicate-api-terms-20261109b | An account needs a card or pre-payment before the services can be used, and each account is limited to 200 concurrent requests. DeepInfra pricing page captured 9 November 2026. The page says pre-paid accounts draw down their balance, invoiced accounts are billed once at the beginning of the following month, a spending limit can be set, and the page invites contacting DeepInfra for more than 200 concurrent requests. Supplier statement, not a hands-on test. Geprüft am 9. Nov. 2026 · DeepInfra: deepinfra-api-pricing-20261109b | The SDK documents a watsonx.ai lightweight engine mode that needs no project or space ID and uses watsonx.ai software authentication. Publisher SDK documentation at captured repository revision, 9 November 2026. Describes the client call only; which engine or deployment an organisation can obtain is not established. No account or model request. Geprüft am 10. Nov. 2026 · IBM watsonx.ai: ibm-watsonx-ai-project | Nicht geprüft | Rate limits start from a per-user default and rise or fall with usage in 15-minute windows, up to 20 times the base allocation; beyond that an Enterprise plan is required. Nebius Token Factory documentation (Rate Limits and Scaling) captured 9 November 2026. The default limits are shown in the account console and were not captured; the baseline in the page example is illustrative. Supplier statement, not a hands-on test. Geprüft am 10. Nov. 2026 · Nebius Token Factory: nebius-token-factory-rates-20261109b |
| Hinweis zum Umfang | Official documentation checked on 18 October 2026; no hands-on test. | Official documentation checked on 18 October 2026; no hands-on test. | Official documentation checked on 18 October 2026; no hands-on test. | Supplier-maintained Python client documentation captured 3 November 2026. Client scope only; no current pricing, availability, residency, model quality or service-level guarantee inferred. No account or inference request. | Supplier-maintained Python client documentation captured 3 November 2026. Client scope only; no current pricing, availability, residency, model quality or service-level guarantee inferred. No account or inference request. | Publisher SDK README captured 3 November 2026; describes that SDK and API examples, not a live service test. Current models, prices, availability and service terms require confirmation. | Publisher SDK README captured 3 November 2026; describes that SDK and API examples, not a live service test. Current models, prices, availability and service terms require confirmation. | Publisher documentation captured 3 November 2026; describes selected documentation or client behavior, not a live service test. Current model availability, prices and service terms require confirmation. | Publisher documentation captured 3 November 2026; describes selected documentation or client behavior, not a live service test. Current model availability, prices and service terms require confirmation. | Publisher integration README at the captured repository revision, 4 November 2026. Client code and hosted service, model and trademark terms are separate. No account, model call, deployment or benchmark. | Publisher integration README at the captured repository revision, 4 November 2026. Client code and hosted service, model and trademark terms are separate. No account, model call, deployment or benchmark. | Publisher integration README at the captured repository revision, 4 November 2026. Client/reference-code licence does not establish hosted-service, model, image or trademark terms. No account, model call, deployment or benchmark. | Publisher integration README at the captured repository revision, 4 November 2026. Client/reference-code licence does not establish hosted-service, model, image or trademark terms. No account, model call, deployment or benchmark. | Integration README at the captured repository revision, 4 November 2026. Client-code licence and hosted service, model and trademark terms are separate. No account, model call, deployment or benchmark. | Integration README at the captured repository revision, 4 November 2026. Client-code licence and hosted service, model and trademark terms are separate. No account, model call, deployment or benchmark. | Publisher SDK documentation at captured repository revision, 4 November 2026. Client-specific prerequisites and routing do not establish model, service, plan or trademark terms. No account or model request. | Publisher SDK documentation at captured repository revision, 4 November 2026. Client-specific prerequisites and routing do not establish model, service, plan or trademark terms. No account or model request. | Publisher cookbook at captured repository revision, 4 November 2026. Example-specific requirements do not establish model, service, plan or trademark terms. No account, inference or application test. |
Im direkten Vergleich