AI visibility report
AI visibility report for Portkey in LLM Observability Evals & Gateways.
Outside the top three on 23 of the 25 prompts buyers actually ask.
Braintrust is cited on 18 of those losses.
Free trial. Setup comes pre-filled for Portkey.
Track Portkey across these prompts daily.
Start free trialStill absent from 96% of tracked prompt responses
Top-3 citations across 125 prompt × platform pairs
Peer Ranking
Key Metrics
Platform Breakdown
ChatGPT
Perplexity
Gemini Search
Google AI Mode
Bing CopilotHow to read this. Portkey appears in 4% of tracked prompt responses. Presence is absolute coverage; share of voice is relative citation share; sentiment measures tone only when the brand appears.
Where Portkey is losing
Prompts where competitors are visible and Portkey is not.
These prompt-level losses are the first prompts to track and repair.
Where Portkey is winning1
Which LLM gateways handle multi-provider fallback and automatic retries while preserving full trace context across the switch?
Avg # 1.0 · 1 platform
Where Portkey is losing5
I'm looking for an LLM observability platform with a great team collaboration workflow — where engineers and PMs can both review trace quality without SQL knowledge.
Competitors on 5 platforms
Track this promptWhich LLM eval platforms support async evaluation at scale without blocking the inference path or adding latency for end users?
Competitors on 5 platforms
Track this promptWhich LLM tracing platforms export trace data to a data warehouse so analysts can run custom eval queries alongside product metrics?
Competitors on 5 platforms
Track this promptWhich LLM observability tools handle PII redaction and data masking in traces for teams with HIPAA or GDPR compliance requirements?
Competitors on 4 platforms
Track this promptWhich LLM observability platforms stay reliable under traffic spikes from batch eval jobs running thousands of LLM calls simultaneously?
Competitors on 4 platforms
Track this prompt
Track Portkey daily before the next report refresh.
Track these gapsResearch dossierCapabilities, use cases, sources, reviews, pricing, and FAQ
Overview
Portkey is a San Francisco-based AI infrastructure company founded in January 2023 by Rohit Agarwal and Ayush Garg. It provides a unified production stack for generative AI teams, combining an AI Gateway, LLM observability, guardrails, prompt management, and enterprise governance in one platform. Portkey routes traffic to 1,600+ large language models across 60+ providers via a single OpenAI-compatible API, processing over 500 billion tokens daily for 24,000+ organizations. Its open-source gateway has accumulated 10.8k+ GitHub stars. Recognized as a 2025 Gartner Cool Vendor in LLM Observability and top-rated on G2, Portkey raised a $15M Series A in February 2026. In April 2026, Palo Alto Networks announced its intent to acquire Portkey to serve as the AI Gateway for the Prisma AIRS security platform.
Portkey is a full-stack LLMOps platform serving as a unified control plane for production AI. It offers an AI Gateway (routing, fallbacks, load balancing, and semantic caching across 1,600+ LLMs), real-time observability (logs, traces, cost tracking, and 40+ metrics), guardrails (PII redaction, content filtering, and prompt injection prevention), a Prompt Engineering Studio (versioning, deployment, and playground), an MCP Gateway, and enterprise governance (RBAC, SSO, audit logs, and budget controls). Deployable as managed SaaS, hybrid, or fully self-hosted with a 3-line code integration claim and 0.999% uptime SLA.
Key Facts
- Founded
- 2023
- HQ
- San Francisco, CA, USA
- Founders
- Rohit Agarwal, Ayush Garg
- Employees
- 11-50
- Funding
- ~$18M
- ARR
- ~$5M (as of June 2024)
- Customers
- 24,000+ organizations
- Status
- Acquisition by Palo Alto Networks (NASDAQ: PANW) pending, an
Target users
Key Capabilities10
- Unified AI Gateway routing to 1,600+ LLMs across 60+ providers via a single OpenAI-compatible API
- Real-time LLM observability: logs, traces, 40+ metrics, cost attribution, and custom metadata
- Automatic fallbacks, load balancing, retries, and semantic caching for production reliability
- Guardrails engine: PII redaction, content filtering, prompt injection prevention, and LLM-as-judge evaluations
- Prompt Engineering Studio: versioning, testing, variable management, playground, and API deployment
- Enterprise AI governance: RBAC, SSO (Okta/SAML), granular budget and rate limits, and org-wide audit logs
- MCP Gateway for centralized authentication and observability of Model Context Protocol servers
- Virtual key management and multi-tenant workspace isolation
- OpenTelemetry-compliant tracing with open-source MIT-licensed gateway (10.8k+ GitHub stars)
- Open-source LLM pricing database powering cost attribution across 200+ enterprises
Key Use Cases8
- Production LLM application reliability and uptime management across multiple providers
- Enterprise AI cost attribution and budget governance across teams and projects
- Multi-provider LLM routing, fallback, and load balancing for high-availability AI
- Agentic AI governance and observability for autonomous multi-step agent workflows
- Prompt versioning, testing, and deployment across LLM providers
- Security compliance and PII protection for regulated industry AI deployments
- Centralized AI access management and key governance for large engineering organizations
- MCP server authentication and observability for enterprise AI tool ecosystems
Portkey customer outcomes
25+ GenAI use cases managed at 30M policies/month
Used Portkey to manage 25+ GenAI use cases processing 30 million insurance policies per month, gaining visibility into prompt management, per-use-case cost tracking, and API key governance.
Thousands of dollars saved via semantic caching
Deployed Portkey in GitHub CI/CD workflows to cache LLM test runs, eliminating repeated token spend on unchanged tests while maintaining production performance quality.
Replaced a fragmented patchwork of S3 logs, DB queries, and manual visualization with Portkey's unified UI, enabling engineers to identify, debug, and rebuild AI agents with confidence.
10,000+ daily queries managed
Used Portkey's analytics and semantic caching to manage cost, latency, and rate limiting for their AI-first employee help desk (Albus) handling over 10,000 daily queries across OpenAI.
Recent Trend
How AI describes Portkey3
...-provider fallback | Automatic retries | Trace continuity across fallback | Best fit | | --- | --- | --- | --- | --- | | Portkey | ✅ | ✅ | ✅ Explicit Trace ID across all attempts | Best overall for th...
Which LLM gateways handle multi-provider fallback and automatic retries while preserving full trace context across the switch?
...Simplest startup choice | | LiteLLM | ⭐⭐⭐⭐ | ✅ | ✅ | Most mature/flexible | | Helicone | ⭐⭐⭐⭐⭐ | ✅ Excellent | ✅ | Observability-first | | Portkey | ⭐⭐⭐ | ⚠️ Mostly hosted control plane | ✅ Gateway | Teams wanting polished SaaS UX | ### 1\.
I'm evaluating LLM gateway solutions for a startup — which ones have the simplest self-hosted setup with a working UI on day one?
...| Circuit breaker | Multi-provider failover | Enterprise fit | My take | | --- | --- | --- | --- | --- | --- | | 1 | Portkey | Yes | Excellent | ★★★★★ | Best turnkey LLM-specific reliability layer | | 2 | Kong AI Gateway | Yes*...
What are the most production-hardened LLM gateway options for an enterprise team needing 99.9% uptime with circuit-breaker support?
Most cited sources8
Alternatives in LLM Observability Evals & Gateways6
Portkey positions itself as a full-stack 'unified control plane for production AI,' differentiating from point-solution observability tools by bundling an AI Gateway, real-time observability, guardrails, prompt management, and enterprise governance into a single platform.
- Its open-source gateway (10.8k+ GitHub stars) and claimed 3-line integration lower friction versus self-hosted alternatives like LiteLLM.
- Against standalone eval-focused tools such as Braintrust or Galileo, Portkey emphasizes production reliability and cost governance over evaluation depth.
- The April 2026 announced acquisition by Palo Alto Networks positions Portkey's gateway as the foundational AI security layer within the Prisma AIRS platform.
Reviews
Praised
- Easy 3-line integration with minimal code changes to existing stack
- Intuitive observability dashboard and analytics
- Effective LLM cost reduction via caching and intelligent routing
- Unified multi-provider API and automatic fallback management
- Responsive and knowledgeable customer support
- Fast time-to-value with immediate production monitoring
- Reliable fallback and retry logic for production uptime
- Strong RBAC and governance features for enterprise teams
Criticized
- Steep learning curve for teams new to LLMOps
- Documentation gaps in advanced and air-gapped configurations
- Price tracking incomplete or requiring manual updates for some models
- Advanced analytics and visualization still maturing vs. older enterprise tools
- Log retention limits on lower tiers can be restrictive for compliance use cases
- Feature breadth can feel overwhelming for newcomers
Users on G2 and Gartner Peer Insights consistently praise Portkey's ease of integration, intuitive observability dashboard, responsive customer support, and effectiveness at reducing LLM costs through caching and intelligent routing. Enterprises cite the ability to unify multi-provider logging and cost attribution as a major workflow improvement. Criticisms center on a learning curve for teams new to LLMOps, documentation gaps for advanced configurations, incomplete price tracking for some models in air-gapped setups, and advanced analytics features still maturing. Portkey holds a 4.6/5 rating on G2 and was recognized as a 2025 Gartner Cool Vendor in LLM Observability.
Pricing
Portkey offers four tiers. Developer (Free): 10,000 recorded logs/month, 3-day log retention, community support.
- Production
$49/month, 100,000 recorded logs/month with $9 per additional 100k requests (up to 3M), 30-day retention, guardrails, RBAC, semantic caching, and production support.
- Enterprise
Custom pricing, 10M+ recorded logs/month, custom retention, SSO, granular budgets and rate limits, private cloud/VPC deployment, SOC2 Type 2, GDPR, and HIPAA compliance, custom BAAs, and dedicated onboarding. A self-hosted open-source tier (MIT license) is also available with no managed log limits. Pricing is log-volume-based rather than request-based.
Limitations
- Log retention is capped at 30 days on the Production plan; custom retention requires Enterprise contracts estimated at $5,000–$10,000+/month.
- The Production plan limits recorded logs to 100k/month with $9 overage charges per 100k additional requests.
- Price tracking does not work universally across all models, and air-gapped setups require manual pricing updates.
- Advanced analytics and visualization features are still maturing relative to established enterprise observability platforms.
- The platform can feel complex and overwhelming for teams new to LLMOps.
- Documentation quality receives mixed feedback in user reviews, particularly for advanced configurations.
Frequently asked questions
Topic coverageCoverage by buyer topic
Topic Coverage
Prompt-Level Results
| Prompt | Gemini Search | Google AI Mode | Perplexity | ChatGPT | Bing Copilot |
|---|---|---|---|---|---|
Capability1/5 cited (20%) | |||||
Which LLM observability tools handle PII redaction and data masking in traces for teams with HIPAA or GDPR compliance requirements? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
Which LLM evaluation platforms support custom rubric-based scoring for domain-specific correctness beyond generic faithfulness metrics? | Neither your brand nor a competitor was cited | Neither your brand nor a competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
Looking for an eval platform that supports automated safety and toxicity scoring on LLM outputs at scale — what are my options? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
What platforms support end-to-end tracing of multi-agent pipelines including tool calls, retrieval steps, and sub-agent spawning? | A competitor was cited | Neither your brand nor a competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited |
Which LLM gateways handle multi-provider fallback and automatic retries while preserving full trace context across the switch? | Neither your brand nor a competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited | Your brand and a competitor were cited | Neither your brand nor a competitor was cited |
Developer Experience0/5 cited (0%) | |||||
What LLM observability tools do ML engineering teams typically use to annotate and review production traces for quality feedback? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
I'm looking for an LLM observability platform with a great team collaboration workflow — where engineers and PMs can both review trace quality without SQL knowledge. | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited |
Which LLM tracing platforms make it easiest to replay a failed multi-step agent run and pinpoint exactly where reasoning went wrong? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
Which LLM eval platforms have the best prompt playground experience for iterating on system prompts against a saved test dataset? | Neither your brand nor a competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
Which LLM gateway tools give developers the best real-time cost and token usage visibility across multiple LLM providers during development? | Neither your brand nor a competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
Integrations & Ecosystem1/5 cited (20%) | |||||
Which LLM observability tools work with OpenTelemetry-compatible backends so we can consolidate LLM traces alongside existing service traces? | A competitor was cited | Neither your brand nor a competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
Which LLM observability platforms integrate natively with the most popular agent frameworks so traces appear automatically with no manual instrumentation? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited |
I'm evaluating LLM eval platforms — which ones integrate with version control to tie prompt regressions back to specific code or config changes? | Neither your brand nor a competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited |
Which LLM tracing platforms export trace data to a data warehouse so analysts can run custom eval queries alongside product metrics? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited |
What LLM gateway tools integrate best with secret managers and internal auth systems for enterprise teams rolling out to multiple product teams? | Neither your brand nor a competitor was cited | Neither your brand nor a competitor was cited | Your brand and a competitor were cited | Your brand and a competitor were cited | Neither your brand nor a competitor was cited |
Performance & Reliability1/5 cited (20%) | |||||
Which LLM observability platforms stay reliable under traffic spikes from batch eval jobs running thousands of LLM calls simultaneously? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited |
What are the most production-hardened LLM gateway options for an enterprise team needing 99.9% uptime with circuit-breaker support? | Neither your brand nor a competitor was cited | A competitor was cited | Your brand and a competitor were cited | Neither your brand nor a competitor was cited | Neither your brand nor a competitor was cited |
What LLM tracing platforms handle high-throughput production workloads — millions of traces per day — without degrading query performance? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
Which LLM eval platforms support async evaluation at scale without blocking the inference path or adding latency for end users? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited |
Which LLM gateways add the least latency overhead when routing between LLM providers — safe to use in production for sub-500ms SLAs? | Neither your brand nor a competitor was cited | Neither your brand nor a competitor was cited | Neither your brand nor a competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
Setup & First Run1/5 cited (20%) | |||||
Which LLM observability platforms can a small team get running against a production RAG pipeline in under a day? | A competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited | A competitor was cited |
What are the best OpenTelemetry-compatible tracing backends for LLM apps that work out of the box without custom span parsing? | A competitor was cited | Neither your brand nor a competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
What's the fastest LLM tracing platform for instrumenting a Python-based agent framework without rewriting existing code? | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited |
I'm evaluating LLM gateway solutions for a startup — which ones have the simplest self-hosted setup with a working UI on day one? | Neither your brand nor a competitor was cited | A competitor was cited | Neither your brand nor a competitor was cited | Your brand and a competitor were cited | Neither your brand nor a competitor was cited |
Which evaluation platforms for LLM outputs are easiest to plug into an existing CI pipeline for a five-engineer team? | Neither your brand nor a competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited | A competitor was cited |
Turn this matrix into daily prompt monitoring.
Track prompt changesVertical Ranking
| # | Brand | PresencePres. | Share of VoiceSoV | DocsDocs | BlogBlog | MentionsMent. | Avg PosPos | Sentiment |
|---|---|---|---|---|---|---|---|---|
| 1 | Braintrust | 32.0% | 21.8% | 7.2% | 0.0% | 44.0% | #3.3 | +0.51 |
| 2 | LangChain | 26.4% | 18.8% | 8.8% | 0.0% | 48.0% | #3.3 | +0.42 |
| 3 | Langfuse | 21.6% | 14.3% | 8.0% | 4.0% | 55.2% | #3.1 | +0.59 |
| 4 | Confident AI | 18.4% | 14.7% | 0.0% | 0.0% | 15.2% | #4.9 | +0.43 |
| 5 | Arize AI | 16.0% | 10.9% | 0.0% | 4.8% | 15.2% | #4.4 | +0.51 |
| 6 | Galileo | 12.0% | 6.8% | 0.0% | 12.0% | 10.4% | #2.8 | +0.40 |
| 7 | LiteLLM | 5.6% | 4.1% | 3.2% | 0.0% | 0.0% | #3.4 | +0.52 |
| 8 | Traceloop | 4.0% | 3.8% | 0.0% | 2.4% | 6.4% | #3.6 | +0.25 |
| 9 | Portkey | 4.0% | 3.4% | 0.8% | 0.0% | 21.6% | #5.3 | +0.45 |
| 10 | Helicone | 2.4% | 1.5% | 1.6% | 0.8% | 25.6% | #5.8 | +0.83 |
| 11 | Patronus AI | 0.0% | 0.0% | 0.0% | 0.0% | 0.8% | — | — |
Turn this into your team dashboard
Sign up to unlock project-level analytics, daily tracking, actionable insights, custom prompt configurations, adoption tracking, AI traffic analytics and more.
Free trial. Setup comes pre-filled from this report.