Respan

What is Respan?

Respan (formerly known as Keywords AI) is an AI router that puts one OpenAI-compatible API in front of more than 1,000 models, with observability and evaluation built into the same platform. You set a fallback chain once, and requests move to the next model when a provider errors or rate-limits; failed calls retry, traffic can be load-balanced, and identical repeat requests are answered from cache. You can set firm spending caps for each key, each customer or the entire org. Each request is recorded as a span carrying its latency and cost, monitors alert through Slack, email or a webhook, and evals score output on test sets and on live traffic. Prompts are versioned and deployed without a code change. Respan has a Free plan at $0, a Team plan at $199 per month, and custom Enterprise packages with SSO, RBAC and an SLA.

Use cases

• Route LLM requests across many providers
• Keep agents online during provider outages
• Cap LLM spend per customer or API key
• Trace agent runs step by step
• Score output quality on live traffic
• Version and deploy prompts without code

Who it's for

• AI engineers
• Product engineering teams
• Platform teams
• Enterprise IT and security

Why use Respan

• Gateway, tracing and evals share one platform, so each metric links back to the exact request behind it.
• Switching models is a one-word change through the OpenAI-compatible API, so provider moves do not mean a rewrite.
• Hard spend caps block requests at the ceiling instead of only sending a warning.
• The Free plan includes unlimited seats and the full gateway, which lets a team trial it before paying.
• SOC 2 Type II, HIPAA with a BAA, GDPR and ISO 27001 are listed for procurement reviews.

Respan features

• One OpenAI-compatible API for 1,000+ models
• Automatic fallbacks, retries and load balancing
• Exact-match caching with configurable TTL
• Hard spend limits per key, customer or org
• Nested traces for LLM calls, tool runs and agent turns
• LLM-judge, code and human evaluators
• Datasets built from production logs
• Monitors for cost, errors, latency and tokens
• Prompt versioning and deployment without code changes
• Task-aware model routing with Respan Router (beta)
• Key vault with bring-your-own-key support

Respan integrations

• Slack
• LangChain
• Claude Code
• Cursor
• Pinecone
• Qdrant
• Chroma
• Milvus
• LlamaIndex
• Supabase
• Mem0
• BigQuery
• Snowflake

How to get started with Respan

1. Create a Respan account and generate an API key, or bring your own provider keys through the key vault.
2. Point your existing OpenAI-compatible SDK at the Respan endpoint and pick a model.
3. Set a fallback chain, caching rules and spend limits at the org level or per request.
4. Open traces for each LLM call, tool run and agent turn, and add monitors that alert on cost, errors or latency.
5. Build datasets from production logs, run evaluators on test sets and live traffic, and ship the prompt or model version that scores best.

Respan limitations

• Routing through a gateway adds roughly 30-40ms of latency
• Free plan includes 100k logs, 1k scores and 2 evaluators
• Free plan keeps data for 7 days
• HIPAA compliance is an add-on at $249 / mo on the pricing page

Respan pricing

Yes, Respan has a free plan. Paid plans start at $199 per month, according to the official pricing page on 24 Oct 2026.

Respan plans

• Free

  $0

  No credit card required

  • 100k logs per month
  • 1k scores, 2 evaluators, 5 datasets, 5 prompts
  • Unlimited seats
  • Full gateway with fallbacks, caching, and spend limits

• Team

  $199 per month

  • Unlimited datasets, evaluators, and prompts
  • 10k scores, then $1 / 1k scores
  • 100k logs, then $8 / 100k logs
  • 30-day retention
  • 5 members included, then $15 / member
  • 99.9% uptime SLA and private Slack channel

• Enterprise

  Custom

  • Custom packages and volume discount
  • SSO and RBAC
  • Dedicated support engineer
  • 99.99% uptime SLA
  • HIPAA BAA

Plans as listed on the official pricing page on 24 Oct 2026. Prices can change and may vary by country or exclude tax. Check the official pricing page.

Respan alternatives

The main Respan alternatives listed on Debuly are OpenRouter, Portkey, LiteLLM, Helicone, Langfuse and Braintrust. The table compares pricing and platforms; order is A–Z.

1. 

   Braintrust

   AI observability and evaluation platform for tracing LLM apps and agents, scoring outputs and catching regressions before release.

   Developer platformsCoding
   FreemiumVisit Braintrust (official site)
2. 

   Helicone

   Open-source LLM observability and AI gateway that logs requests, tracks cost and latency, and routes to 100+ models through one API.

   Developer platformsCoding
   FreemiumVisit Helicone (official site)
3. 

   Langfuse

   Open-source LLM observability and evaluation platform for tracing agents, managing prompts and scoring outputs.

   Developer platformsAutomation and agentsCoding
   FreemiumVisit Langfuse (official site)
4. 

   LiteLLM

   Open-source AI gateway and Python SDK that gives one OpenAI-format API to 100+ LLM providers, with keys, budgets and cost tracking.

   Developer platformsCodingAutomation and agents
   Open sourceVisit LiteLLM (official site)
5. 

   OpenRouter

   Unified OpenAI-compatible API for 500+ AI models with provider fallbacks, pay-as-you-go credits and a free tier

   Developer platformsChat assistantsAutomation and agents
   FreemiumVisit OpenRouter (official site)
6. 

   Portkey

   AI gateway with observability, guardrails, caching and prompt management for teams running LLM apps in production.

   Developer platformsAutomation and agents
   FreemiumVisit Portkey (official site)

Respan FAQ

Sources & verification

Facts on this page come from the official sources below, each with the date we read it. Written by Debuly; the vendor did not review this page.

Image credits

• Logo: from respan.ai · fetched 24 Oct 2026 · Vendor trademark, used only to identify the product; removed on request
• Cover: none collected yet, so a generated tile is shown.

Spot an error? Report an issue with this listing. We also remove a logo or image on the owner's request.