tools
Dataiku LLM Mesh
Dataiku LLM Mesh connects applications to multiple hosted and self-hosted LLMs with centralized routing, governance, monitoring, and cost controls.
In inglese
Dataiku LLM Mesh provides a governed gateway between applications and LLM providers. Teams can connect hosted APIs or self-hosted models, switch providers without rewriting applications, route requests, and apply access, moderation, and audit policies.
It is used by enterprise AI, data science, IT, and application teams building chatbots, retrieval-augmented generation systems, agents, and other LLM workflows. The product is part of Dataiku and requires a Dataiku deployment; model-provider charges and any applicable Dataiku subscription costs are separate.
Features
- Connect to hosted LLM APIs and locally running Hugging Face models
- Route requests across providers and switch models without rewriting applications
- Apply centralized access controls, usage limits, moderation, and audit policies
- Monitor LLM usage, cost, latency, and round-trip performance
- Cache frequent responses to reduce repeated requests and spend
- Use completion, embedding, streaming, multimodal, and image-generation APIs
- Create persistent multi-turn conversations and query knowledge banks
- Access the LLM Mesh through Python and REST APIs
Use cases
- Route prompts to different models based on cost, performance, or compliance
- Govern access to approved LLM providers across enterprise applications
- Monitor token usage, latency, and provider spend
- Build retrieval-augmented chatbots using internal knowledge banks
- Serve multimodal and image-generation workflows through one API
- Connect Dataiku agents and applications to multiple model providers
Pros
Cons
Latest updates
- New feature: Leverage existing external Vector Stores (14.5.0)
Leverage existing external Vector Stores
- New feature: Automatic throttling for LLM connections (14.0.0)
Automatic throttling for LLM connections
- New feature: Enhanced documents embedding (14.0.0)
Enhanced documents embedding
- New Feature: Document embedding (13.4.0)
Document embedding
- New feature: LLM cost blocking & alerting (13.4.0)
LLM cost blocking & alerting
Capabilities
- Choice of models — “Switch or mix models based on cost, performance, and compliance needs without breaking your applications.” source
- Self-hosted — “Connect to a broad ecosystem of LLM providers and self-hosted models through a secure, agnostic gateway.” source
- API — “Send completion and embedding queries to all LLMs supported by the LLM Mesh” source
- Search over your data — “Interact with knowledge banks, and perform semantic search” source
- Traces and evaluates — “Monitor LLM usage and round-trip performance in real time, cache frequent responses for efficiency, and pick the right model for each task, avoiding runaway spend while meeting SLAs.” source
Security
- SOC 2 Type II — “Dataiku has completed SOC 1 Type II and SOC 2 Type II assessments to assure our customers that internal controls are in place to protect their data.” source
- SOC 2 — “Dataiku has completed SOC 1 Type II and SOC 2 Type II assessments to assure our customers that internal controls are in place to protect their data.” source
- ISO 27001 — “Dataiku is ISO 27001:2022-certified, demonstrating that it has implemented and maintained an Information Security Management System (ISMS).” source
- ISO 42001 — “Dataiku is ISO 42001:2023-certified for the implementation and governance of its Artificial Intelligence Management System (AIMS).” source
- GDPR — “Dataiku is compliant with global data protection regulations, such as the GDPR and the CCPA.” source
- HIPAA — “To support the compliance programs for our Healthcare/ Life Sciences customers, Dataiku has voluntarily extended its Trust program to include a HIPAA (Health Insurance Portability and Accountability Act) compliance report.” source
Pricing
- Prices checked
- 2026-09-25