Question 1

What AI providers does Axemere support?

Accepted Answer

Axemere supports OpenAI (GPT-4, GPT-4o, o1), Anthropic (Claude 3.5 Sonnet, Claude 3 Opus, Haiku), Google Gemini (1.5 Pro, 1.5 Flash), Cohere (Command R+), and Azure OpenAI. New providers are added regularly; check the documentation for the current list.

Question 2

Is the self-hosted gateway really free?

Accepted Answer

Yes. The self-hosted gateway is free, open to use, and has no usage limits. There is no account required. Download it, configure a .env file with your API keys, and run it. You pay your AI providers directly; Axemere charges nothing for the self-hosted gateway itself.

Question 3

How does Axemere keep my API keys secure?

Accepted Answer

Your API keys are stored inside the gateway process and are never exposed to your application code. Applications send requests through the gateway using a gateway credential (a short-lived token or a local request), and the gateway attaches the actual provider key before forwarding the request. Keys are encrypted at rest in the Axemere control plane and are never returned in API responses.

Question 4

What is the difference between the self-hosted gateway and the managed gateway?

Accepted Answer

The self-hosted gateway runs in your own infrastructure: on your laptop, a VM, Docker, or Kubernetes. You are responsible for operating, updating, and scaling it. The managed gateway is hosted and operated by Axemere; you get a gateway endpoint without running any infrastructure. Both support the same AI providers and the same drop-in proxy interface.

Question 5

Does Axemere store my AI request and response content?

Accepted Answer

No. Axemere does not store, view, or log the content of prompts or responses, under any circumstances. For the self-hosted gateway, request and response content never leaves your infrastructure. For the managed gateway, requests route through Axemere's infrastructure to reach the provider, but the content itself is held in memory only for the duration of policy evaluation and dispatch, then discarded. What Axemere retains is execution-record metadata (org/workload attribution, model, token counts, cost, policy decision, timestamps) plus a SHA-256 hash of the request/response for tamper-evident audit purposes. The raw text is never written to any database or log.

Question 6

Can I switch AI providers without changing my application code?

Accepted Answer

Yes. The gateway exposes a single base URL per provider type. When you want to switch providers, you update the gateway configuration, not your application. Applications that use the OpenAI SDK continue working unchanged; the gateway routes the request to whichever provider you configure.

Question 7

How do budget caps work?

Accepted Answer

You set a spending limit per credential or project in the gateway configuration. The gateway tracks token usage in real time and blocks requests once the threshold is hit, before the request reaches the provider. This prevents overage charges at billing time. Budget resets are configurable (daily, weekly, monthly).

Question 8

What is the Control Plane and do I need it?

Accepted Answer

The Control Plane is the cloud management layer for your gateway fleet. It provides cross-gateway policy enforcement, spend attribution by project and team, approval workflows, and a cryptographic audit trail. Individual developers using a single self-hosted gateway do not need the Control Plane. Teams and enterprises that want centralized governance across multiple gateways or require audit-grade records typically do.

Question 9

How do I integrate Axemere with LangChain?

Accepted Answer

How do I integrate Axemere with LangChain?

Question 10

Does Axemere work with frameworks other than LangChain?

Accepted Answer

Yes. Because the gateway is a drop-in OpenAI-compatible proxy, it works with any framework or SDK that supports a configurable base URL: LangChain, LlamaIndex, Vercel AI SDK, Haystack, CrewAI, AutoGen, and direct HTTP calls. You set the base URL to your gateway endpoint and the rest of your code is unchanged.

Question 11

What is included in the Growth Pack managed gateway plan?

Accepted Answer

Growth Pack ($25/month per pack, 30-day free trial, no credit card required) is a single package purchased in quantity as usage grows: buy more packs at a volume discount instead of jumping to a fixed second tier. Each pack includes a hosted gateway with unlimited team members, unlimited requests (governed by an abuse-prevention rate limit, not billed), 10M input tokens and 25M output tokens per pack, redundancy across availability zones, full execution record history, and the full Axemere console. It is designed for individual developers and production teams alike.

Question 12

Can I self-host the gateway and also use the Control Plane?

Accepted Answer

Yes. The Control Plane supports both managed and self-hosted gateways. You can run gateways in your own infrastructure and still get centralized policy management, cross-gateway spend attribution, and audit trails through the Control Plane. This is the recommended architecture for organizations with strict data residency requirements.

Question 13

Is there an audit trail for AI usage?

Accepted Answer

Yes. The gateway records every execution: model, provider, token counts, latency, status code, and the credential that made the request. The Control Plane aggregates these records across gateways and makes them queryable. The Compliance package provides cryptographically signed, tamper-evident audit exports suitable for regulatory submissions.

Question 14

How does Axemere handle multi-model routing?

Accepted Answer

You can configure routing rules in the gateway that send requests to different providers based on model name, cost, latency, or availability. For example, you can route Claude requests to Anthropic by default but fall back to a fine-tuned Azure model if the primary is unavailable. Routing rules are defined in the gateway configuration file and do not require application changes.

Question 15

Can I run Axemere on Kubernetes?

Accepted Answer

Yes. The self-hosted gateway ships as a container image and includes Helm chart support. You can deploy it as a standard Kubernetes service with horizontal scaling, health checks, and persistent configuration via ConfigMaps or Secrets. Documentation for Kubernetes deployment is in the quick start guide.

Question 16

What compliance certifications does Axemere have?

Accepted Answer

Axemere is working toward SOC 2 Type II certification. All traffic is encrypted in transit (mTLS between every internal service hop). At rest, stored credential secrets (your AI provider API keys) get an additional application-layer encryption step (AES-256-GCM) on top of the database's own disk-level encryption; all other platform data (execution records, usage, and cost data) relies on that same standard managed-database encryption at rest. See the Security Overview docs for the full breakdown. The Compliance package provides audit-grade execution records and export capabilities designed to support GDPR, HIPAA, and financial services regulatory requirements. Contact sales for specific compliance questions.

Question 17

How does spend attribution work across teams?

Accepted Answer

You assign credentials to projects, and projects to teams. Every execution record carries the project and credential that made the request. The Control Plane aggregates this data and surfaces spend breakdowns by project, team, model, and provider in the analytics dashboard. The Advanced Analytics package adds cross-gateway reporting and custom attribution dimensions.

Question 18

Does Axemere work with the LangChain demo referenced in your documentation?

Accepted Answer

Does Axemere work with the LangChain demo referenced in your documentation?

Question 19

What is the fastest way to get started with Axemere?

Accepted Answer

What is the fastest way to get started with Axemere?

Frequently Asked Questions