Is Conifer an alternative to Portkey?

For the gateway, yes. Both put one API in front of models from many providers, with fallbacks, spending limits and cost records. Portkey also offers guardrails, prompt management and hybrid or self-hosted deployment. Conifer is hosted and also sells model usage from a prepaid balance.

Who owns Portkey now?

Palo Alto Networks acquired Portkey in 2026. Portkey’s site now says Portkey is Prisma AIRS AI Gateway, which Palo Alto Networks announced as generally available in July 2026.

Do I need provider accounts to use Conifer?

No. One Conifer key covers every model in the catalog, paid from a prepaid balance. You can also put your own Azure OpenAI resource behind the same key, and Conifer adds no fee on that usage.

Does Conifer store my prompts?

The gateway does not store the prompt of a standard request. It holds a response for ten minutes so that a retried request is not charged twice, and keeps request details such as the model, token counts and cost for billing.

Comparison

Conifer and Portkey compared

Both put one gateway in front of AI models from many providers. This page sets out how each one works, using each company’s own published documentation.

Summary

Portkey is a gateway and operations platform with observability, guardrails, prompt management and governance in one product. It runs on Portkey’s cloud, in a hybrid setup inside your environment, or from its open-source gateway, and it connects to provider accounts you hold. Conifer is a hosted gateway that also sells model usage from a prepaid balance, serves a named model only as named, and reports the exact cost of every request.

Portkey is now part of Palo Alto Networks and is offered as Prisma AIRS AI Gateway. Portkey details were read from portkey.ai, its documentation and Palo Alto Networks on 7 November 2026. Plans and prices change; check the linked pages before you decide.

Side by side

Topic	Conifer	Portkey
Where it runs	A hosted service at one Conifer address; there is nothing to deploy	Portkey’s cloud, or a hybrid setup in which the gateway runs in your environment and Portkey runs the management side. New fully air-gapped deployments are no longer offered
Open source	The Conifer SDK for TypeScript and Python is open source; the gateway is a hosted service	The gateway is open source under the MIT licence and can run locally with one command, in Docker, on Node.js or on Cloudflare Workers
Model access	More than 230 models paid from one prepaid balance, with no provider accounts needed; your own Azure OpenAI resource can sit behind the same key	You connect your own provider accounts and Portkey stores their credentials. Its open-source gateway describes routing to “1600+ language, vision, audio, and image models”
Interfaces	OpenAI Chat Completions, OpenAI Responses and Anthropic Messages, on one key	OpenAI Chat Completions, OpenAI Responses and Anthropic Messages formats, translated between providers
What you pay	Per-token prices published in the catalog, from a prepaid balance, with no subscription. No Conifer fee on usage of your own Azure resource	Plans priced by recorded logs: Developer is free with 10,000 a month; Production is $49 a month with 100,000, then $9 for each further 100,000; Enterprise is custom
A named model	Served as named, or refused. Other models are tried only if you list up to three fallbacks yourself	A config lists providers or models in order, and if the first fails Portkey moves to the next. Fallbacks can be limited to chosen error codes
Routing	Conifer picks the provider for each request and retries the same model elsewhere on failure. The router, in beta, picks a model from a pool you approve	Load balancing across providers or models, conditional routing on rules you define, retries and request timeouts
Cost on each request	Every response carries the amount charged, itemised by token type, and a streamed response ends with the same receipt	Each log records the cost of the request, calculated from Portkey’s pricing data for the models it supports
Spending limits	A cost ceiling per request, a spend cap per key and a monthly limit on the account. A request that could exceed a ceiling is refused before any provider is called	Budgets and rate limits for each workspace on shared provider credentials. The plan comparison lists budgeting and rate limiting with key management on Enterprise
Request logs	The prompt of a standard request is not stored. A response is held for ten minutes so a retry is not charged twice, and request details such as model, tokens and cost are kept for billing	Each log shows the raw request and response. Logs are kept 3 days on Developer, 30 days on Production and for a custom period on Enterprise, and the plan table lists a privacy mode
Beyond the gateway	The docs cover the gateway, the router and the SDKs	Guardrails, prompt management, an MCP gateway, and observability with traces and alerts
Enterprise terms	Team accounts with owner, admin and member roles, per-key model scope and spend caps, set up on request	Enterprise lists SSO, SOC 2 Type 2, GDPR and HIPAA compliance, custom BAAs, VPC hosting and private cloud deployment

Which one fits

Portkey may suit you if: You want guardrails, prompt management and full request logs in the same product as the gateway, the gateway must run inside your own environment, you already buy models directly from each provider, or your security team is standardizing on Palo Alto Networks.
Conifer may suit you if: You want many models on one prepaid balance without an account with each provider, you need each response to state the model that served and the exact amount charged, you want requests refused before they exceed a cost ceiling, or you want automatic model choice limited to models you approve.

Moving from Portkey to Conifer

Address and key: Point an OpenAI or Anthropic client at Conifer and use a Conifer key.
OPENAI_BASE_URL=https://api.conifer.build/v1 OPENAI_API_KEY=$CONIFER_API_KEY
Model names: Use the model names in the Conifer catalog. Names with the lab in front, such as anthropic/claude-opus-5, are also accepted.
Providers and fallbacks: Conifer picks the provider for each request and tries another provider for the same model if one fails, so provider lists are not needed. To try other models, list up to three fallbacks in a request header.
Your own Azure resource: Paste the Azure OpenAI resource address and key once in the console. Callers keep using their Conifer key, and usage is billed to your Azure agreement.
Check first: AWS Bedrock and Google Vertex are not served as your own endpoints today, and the moderations endpoint is not served.

Replace your provider key has the full steps, Bring your own endpoint covers Azure, and Routing covers fallbacks and failover.

Conifer Gateway: One API key and endpoint for every model in the catalog, with the model and cost stated on every response.
What is an LLM gateway?: How a gateway sits between your software and the companies that serve models, and what to check when you choose one.
Conifer vs OpenRouter: The two gateways side by side, from each company's own documentation.
Conifer vs Helicone: The two gateways side by side, from each company's own documentation.
Conifer vs LiteLLM: The two gateways side by side, from each company's own documentation.
All comparisons: Conifer compared with eleven other LLM gateways and routers, from their own documentation.

Questions

Is Conifer an alternative to Portkey?: For the gateway, yes. Both put one API in front of models from many providers, with fallbacks, spending limits and cost records. Portkey also offers guardrails, prompt management and hybrid or self-hosted deployment. Conifer is hosted and also sells model usage from a prepaid balance.
Who owns Portkey now?: Palo Alto Networks acquired Portkey in 2026. Portkey’s site now says Portkey is Prisma AIRS AI Gateway, which Palo Alto Networks announced as generally available in July 2026.
Do I need provider accounts to use Conifer?: No. One Conifer key covers every model in the catalog, paid from a prepaid balance. You can also put your own Azure OpenAI resource behind the same key, and Conifer adds no fee on that usage.
Does Conifer store my prompts?: The gateway does not store the prompt of a standard request. It holds a response for ten minutes so that a retried request is not charged twice, and keeps request details such as the model, token counts and cost for billing.