# Mystic Documentation

> Mystic AI provides model hosting services for startups and enterprise. Get started with low cost pre-made models today like Llama 2, stable diffusion, and many more!

Append .md to any documentation page URL to get its markdown version.

## Guides
- [Getting started](https://docs.mystic.ai/docs/getting-started.md): Learn more about Mystic.
- [FAQ](https://docs.mystic.ai/docs/faq.md): Commonly asked questions
- [Authentication](https://docs.mystic.ai/docs/login.md): How to safely use the Mystic APIs
- [Inference](https://docs.mystic.ai/docs/running-a-pipeline.md): Different options for running your pipeline
- [Pointers](https://docs.mystic.ai/docs/pointers-1.md)
- [Files](https://docs.mystic.ai/docs/files-and-directories.md)
- [Inputs and outputs](https://docs.mystic.ai/docs/inputs-outputs.md): Pipelines implement typed IO fields for predictable inference
- [Quickstart](https://docs.mystic.ai/docs/quickstart.md): Speed run creating and deploying a pipeline
- [Pipeline building](https://docs.mystic.ai/docs/pipeline-building.md): Learn how to build pipelines
- [Streaming](https://docs.mystic.ai/docs/streaming.md): Get output streams from your pipelines
- [Fractional GPUs & GPU sharing](https://docs.mystic.ai/docs/fractional-gpus-gpu-sharing.md): Learn how to lower costs by using a GPU slice instead of a whole GPU!
- [Using the README.md](https://docs.mystic.ai/docs/using-the-readmemd.md): Learn how to use the autogenerated README.md with the pipeline-ai SDK
- [Cold start optimisations](https://docs.mystic.ai/docs/cold-start-optimisations.md)
- [Migration guide](https://docs.mystic.ai/docs/migration-guide.md): How to migrate to v2 (i.e. API V4)
- [Entity objects](https://docs.mystic.ai/docs/entity-objects.md)
- [GPUs and accelerators](https://docs.mystic.ai/docs/gpus-and-accelerators.md)
- [Troubleshooting](https://docs.mystic.ai/docs/troubleshooting.md): What to do when your pipeline is not working
- [Migrating from other frameworks](https://docs.mystic.ai/docs/migrating-from-other-frameworks.md)
- [CI/CD Integration](https://docs.mystic.ai/docs/cicd-integration.md): Upload and deploy pipelines as part of your CI/CD flows
- [Teams](https://docs.mystic.ai/docs/teams.md): Collaborate with others in a team
- [Turbo Registry](https://docs.mystic.ai/docs/turbo-registry.md): The fastest docker cold starts in the world!
- [Overview](https://docs.mystic.ai/docs/overview.md): Mystic's BYOC is the easiest way to deploy ML models on your cloud account
- [Deploying Pipelines](https://docs.mystic.ai/docs/deploying-pipelines.md): Deploy pipelines to a cluster running on your own cloud through Mystic BYOC
- [Scaling configuration](https://docs.mystic.ai/docs/scaling-configuration.md): Control the scaling parameters of your Pipeline to best suit your products needs - scale to 0 or 1k GPUs without writing code or touching a cloud provider.
- [Warmup & Cooldown](https://docs.mystic.ai/docs/warmup-cooldown.md): Notify Mystic that you either want a pipeline to start caching, or you're done with it and want it to scale to 0!
- [Walkthrough video](https://docs.mystic.ai/docs/byoc-walkthrough-video.md): This walkthrough takes you through the full onboarding flow to deploy Stable Diffusion XL in your own cloud through Mystic.
- [Llama 2 chat with vLLM (7B, 13B & multi-gpu 70B)](https://docs.mystic.ai/docs/llama-2-with-vllm-7b-13b-multi-gpu-70b.md): This guide shows how to accelerate Llama 2 inference using the vLLM library for the 7B, 13B and multi GPU vLLM with 70B.
- [Deploy Mistral 7B with vLLM](https://docs.mystic.ai/docs/mistral-ai-7b-vllm-fast-inference-guide.md)
- [Deploy Stable Diffusion](https://docs.mystic.ai/docs/deploy-a-stable-diffusion-pipeline.md): How to deploy a pre-trained HugginFace stable diffusion pipeline to Mystic.
- [How to deploy a Hugging Face model](https://docs.mystic.ai/docs/how-to-deploy-a-huggingface-model.md): We outline how to deploy a HuggingFace pretrained transformers model. For this example, we use a text classification sentiment analysis model.
- [Deploy MusicGen model from AudioCraft](https://docs.mystic.ai/docs/deploy-musicgen-audiocraft-model.md)
- [Deploy Mixtral 8x7B with Exllamav2](https://docs.mystic.ai/docs/deploy-mixtral-8x7b-with-exllamav2.md)
- [How to reduce cold starts in ML models running in production](https://docs.mystic.ai/docs/how-to-reduce-cold-starts-in-ml-models.md): Cold boots take a long time in ML and AI, this guide covers what you can and can't do to help this fix this issue
- [Transcribe YouTube videos](https://docs.mystic.ai/docs/get-youtube-transcripts-using-ml.md): Transcribe YouTube videos to get transcripts with timestamps using Whisper Large v3 from Open AI
- [Deploy Gemma 7B with TensorRT-LLM](https://docs.mystic.ai/docs/deploy-gemma7b-tensorrt-llm.md)
- [Deploy Llama 3 70B with ExLlamav2](https://docs.mystic.ai/docs/deploy-llama-3-70b-with-exllamav2.md)
- [Run ComfyUI as an API](https://docs.mystic.ai/docs/run-comfyui-as-an-api.md): Deploy your ComfyUI workflow as a serverless endpoint with Mystic

## API Reference
- [Index](https://docs.mystic.ai/reference/index_v4_pipelines_get.md): Retrieve a paginated set of pipelines.
- [Create](https://docs.mystic.ai/reference/create_v4_pipelines_post.md): Upload a new pipeline.
- [List Pipelines By Family](https://docs.mystic.ai/reference/list_pipelines_by_family_v4_pipelines_by_family__pipeline_family_name__get.md): Retrieve all pipelines for a given pipeline family. Include private pipelines if the user owns the family.
- [Retrieve By Pointer](https://docs.mystic.ai/reference/retrieve_by_pointer_v4_pipelines_by_pointer__pointer__get.md): Retrieve details for a pipeline by pointer name.  Note that the fields this endpoint returns are dependent on the provided query parameters and will omit any empty fields. Accepted query parameters are: - include_pointers: include any pointers that point to this pipeline
- [Owned Or Used](https://docs.mystic.ai/reference/owned_or_used_v4_pipelines_owned_or_used_get.md): Get a lean list of pipelines that the user has either used or owns.
- [Retrieve](https://docs.mystic.ai/reference/retrieve_v4_pipelines__pipeline_id__get.md): Retrieve details for a pipeline.  Note that the fields this endpoint returns are dependent on the provided query parameters and will omit any empty fields. Accepted query parameters are: - include_pointers: include any pointers that point to this pipeline
- [Delete](https://docs.mystic.ai/reference/delete_v4_pipelines__pipeline_id__delete.md): Delete a pipeline.
- [Patch](https://docs.mystic.ai/reference/patch_v4_pipelines__pipeline_id__patch.md): Update selected fields on a pipeline
- [Clone](https://docs.mystic.ai/reference/clone_v4_pipelines__pipeline_id__clone_post.md)
- [Patch Image](https://docs.mystic.ai/reference/patch_image_v4_pipelines__pipeline_id__image_patch.md): Update the image avatar of a pipeline. The new image url will be available in the response.
- [Retrieve Readme](https://docs.mystic.ai/reference/retrieve_readme_v4_pipelines__pipeline_id__readme_get.md): Retrieve a pipeline's readme.
- [Retrieve Avg Cost](https://docs.mystic.ai/reference/retrieve_avg_cost_v4_pipelines__pipeline_id__avg_cost_get.md): Retrieve a pipeline's average cost per run over the past 1000 runs.
- [Pipeline Warmup](https://docs.mystic.ai/reference/pipeline_warmup_v4_pipelines__pipeline_id__warmup_post.md)
- [Pipeline Cooldown](https://docs.mystic.ai/reference/pipeline_cooldown_v4_pipelines__pipeline_id__cooldown_post.md)
- [Retrieve Scaling Info](https://docs.mystic.ai/reference/retrieve_scaling_info_v4_pipelines__pipeline_id__scaling_get.md): Get real-time scaling information for the given pipeline.
- [Get Logged In User](https://docs.mystic.ai/reference/get_logged_in_user_v4_users_me_get.md): Retrieve the currently logged in user.
- [Get User Public Profile](https://docs.mystic.ai/reference/get_user_public_profile_v4_users__username__get.md): Retrieve a user's public profile
- [Validate](https://docs.mystic.ai/reference/validate_v4_tokens_validate_get.md): Validates a header token or cookie JWT.
- [Index](https://docs.mystic.ai/reference/index_v4_tokens_get.md): Retrieve a Catalyst API Bearer token for an authenticated user.
- [Create](https://docs.mystic.ai/reference/create_v4_tokens_post.md): Create a new Catalyst API Bearer token for an authenticated user.  This token will be associated with a team if a team_id is present in the JWT.
- [Retrieve](https://docs.mystic.ai/reference/retrieve_v4_tokens__token_id__get.md): Retrieve an API Bearer token for an authenticated user.
- [Delete](https://docs.mystic.ai/reference/delete_v4_tokens__token_id__delete.md): Delete an API Bearer token for an authenticated user.
- [Patch](https://docs.mystic.ai/reference/patch_v4_tokens__token_id__patch.md): Update selected fields of an API Bearer token for an authenticated user.
- [Index](https://docs.mystic.ai/reference/index_v4_runs_get.md): Retrieve a paginated set of runs that satisfy query parameters.  Supported query params:  - `pipeline_id`: filter by a given pipeline ID     - e.g. `pipeline_id=pipeline_some-random-string` - `order_by`: order by a given field     - supported fields are `created_at`, `state`, `compute_time_ms`; default         is `created_at:desc`     - e.g. `order_by=compute_time_ms:desc` - include_pointers: include any pointers that point to the pipelines the runs were executed against
- [Submit Run](https://docs.mystic.ai/reference/submit_run_v4_runs_post.md): Submits a new run request
- [Stream Run](https://docs.mystic.ai/reference/stream_run_v4_runs_stream_post.md): Submit a new streaming run request, where the run output is streamed as it becomes available, rather than waiting for all the data to be generated.
- [Retrieve](https://docs.mystic.ai/reference/retrieve_v4_runs__run_id__get.md): Retrieve the details of a run.
- [Retrieve Detailed](https://docs.mystic.ai/reference/retrieve_detailed_v4_runs__run_id__detailed_get.md): Retrieve a more detailed version of a run.
- [Upcoming Invoice](https://docs.mystic.ai/reference/upcoming_invoice_v4_billing_upcoming_invoice_get.md)
- [Index](https://docs.mystic.ai/reference/index_v4_pointers_get.md): Retrieve a paginated set of pipeline pointers that satisfy query parameters.  Users are authorised to query for pointers:  -related to a given pipeline family     `pipeline_name=marjory92/falcon`
- [Create](https://docs.mystic.ai/reference/create_v4_pointers_post.md): Create a pipeline pointer instance.
- [Delete](https://docs.mystic.ai/reference/delete_v4_pointers__pointer__delete.md)
- [Pointers Patch](https://docs.mystic.ai/reference/pointers_patch_v4_pointers__pointer__patch.md): Update a pipeline pointer instance.
- [Index](https://docs.mystic.ai/reference/index_v4_pipeline_families_get.md): List pipeline families. Depending on parameters this can mean all public families, or a user's specific families. Can also be ordered and filtered based on family name.
- [Retrieve](https://docs.mystic.ai/reference/retrieve_v4_pipeline_families__pipeline_family_name__get.md): Retrieve a pipeline family by name with its latest pipeline. If the user owns the family, the latest pipeline may be a private pipeline, otherwise it will only be public.
- [Index](https://docs.mystic.ai/reference/index_v4_files_get.md): Retrieve a paginated set of pipeline file instances.
- [Create](https://docs.mystic.ai/reference/create_v4_files_post.md): Create a pipeline file instance.
- [Retrieve](https://docs.mystic.ai/reference/retrieve_v4_files__file_id__get.md): Retrieve a pipeline file instance.
- [Used Pipelines](https://docs.mystic.ai/reference/used_pipelines_v4_metrics_pipelines_get.md): Retrieve paginated metrics for all pipelines used by a user, optionally filtered by 'search'.
- [Top Pipelines Usage](https://docs.mystic.ai/reference/top_pipelines_usage_v4_metrics_top_pipelines_usage_get.md): Retrieve usage metrics for top 'num_pipelines' (in terms of number of runs) for the provided time period.
- [Overall Usage Summary](https://docs.mystic.ai/reference/overall_usage_summary_v4_metrics_overall_usage_summary_get.md)
- [Get Run Logs](https://docs.mystic.ai/reference/get_run_logs_v4_logs_run__run_id__get.md): Get the pipeline logs for a run
- [Get Pipeline Startup Logs](https://docs.mystic.ai/reference/get_pipeline_startup_logs_v4_logs_pipeline_startup__pipeline_id_or_pointer__get.md): Get the logs for pipeline starting up
- [List Credentials](https://docs.mystic.ai/reference/list_credentials_v4_cloud_credentials_get.md): List all credentials belonging to a user
- [Delete Cloud Credential](https://docs.mystic.ai/reference/delete_cloud_credential_v4_cloud_credentials__credential_id__delete.md)
- [Auth Gcp](https://docs.mystic.ai/reference/auth_gcp_v4_cloud_provider_gcp_auth_get.md)
- [Callback Gcp](https://docs.mystic.ai/reference/callback_gcp_v4_cloud_provider_gcp_callback_get.md)
- [Create Gcp Service Account](https://docs.mystic.ai/reference/create_gcp_service_account_v4_cloud_provider_gcp_service_accounts_post.md)
- [List Projects](https://docs.mystic.ai/reference/list_projects_v4_cloud_provider_gcp_projects_get.md): Get all projects on a user's cloud account
- [List Regions](https://docs.mystic.ai/reference/list_regions_v4_cloud_provider_gcp_projects__project_id__regions_get.md): Get all the available regions under a project
- [Auth Aws](https://docs.mystic.ai/reference/auth_aws_v4_cloud_provider_aws_auth_post.md)
- [Auth Azure](https://docs.mystic.ai/reference/auth_azure_v4_cloud_provider_azure_auth_post.md)
- [List Clusters](https://docs.mystic.ai/reference/list_clusters_v4_clusters_get.md): List all clusters belonging to a user
- [Create Cluster](https://docs.mystic.ai/reference/create_cluster_v4_clusters_post.md)
- [List Node Pools](https://docs.mystic.ai/reference/list_node_pools_v4_clusters__cluster_id__node_pools_get.md): List node pools for a given cluster
- [Create Node Pool](https://docs.mystic.ai/reference/create_node_pool_v4_clusters__cluster_id__node_pools_post.md)
- [List Pipelines](https://docs.mystic.ai/reference/list_pipelines_v4_clusters__cluster_id__pipelines_get.md): List pipelines deployed to a given cluster
- [Delete Cluster](https://docs.mystic.ai/reference/delete_cluster_v4_clusters__cluster_id__delete.md)
- [Delete Node Pool](https://docs.mystic.ai/reference/delete_node_pool_v4_clusters__cluster_id__node_pools__node_pool_name__delete.md)
- [Set Pod Info](https://docs.mystic.ai/reference/set_pod_info_v4_resources_post.md)
- [Index](https://docs.mystic.ai/reference/index_v4_scaling_configs_get.md): Retrieve a paginated set of scaling configurations
- [Create](https://docs.mystic.ai/reference/create_v4_scaling_configs_post.md): Create a scaling configuration for pipelines
- [Get Pipeline List](https://docs.mystic.ai/reference/get_pipeline_list_v4_scaling_configs_pipelines__name__get.md): Get a list of pipelines scoped to the user which use a scaling configuration
- [Get By Name](https://docs.mystic.ai/reference/get_by_name_v4_scaling_configs__name__get.md): Retrieve a scaling configuration
- [Delete](https://docs.mystic.ai/reference/delete_v4_scaling_configs__name__delete.md): Delete a scaling configuration
- [Patch](https://docs.mystic.ai/reference/patch_v4_scaling_configs__name__patch.md): Patch a scaling configuration
- [Submit Run](https://docs.mystic.ai/reference/submit_run_v3_runs_post.md)
- [Upload File](https://docs.mystic.ai/reference/upload_file_v3_pipeline_files_post.md): Create a pipeline file instance.

## Changelog
- [Welcome to Mystic](https://docs.mystic.ai/changelog/welcome-to-mystic.md)