Verify before you commit
Use this page to sanity-check Replicate quickly. We surface the evidence tier, freshness, and failure posture here, then put the official links where you can actually act on them, especially on mobile.
Evidence
Assessed
Docs reviewed · Mar 6, 2026
Freshness
Updated 2026-03-06T22:21:51.113+00:00
Mar 6, 2026
Failures
Clear
No active failures listed
| Dimension | Score | Bar |
|---|---|---|
| Execution Score Measures reliability, idempotency, error ergonomics, latency distribution, and schema stability. | 8.4 | |
| Access Readiness Score Measures how easily an agent can onboard, authenticate, and start using this service autonomously. | 8.0 | |
| Aggregate AN Score Composite score: 70% execution + 30% access readiness. | 8.3 | |
Self-serve, pay-per-prediction. Usage-based. Credit card via web. Simple pricing.
API tokens, basic team features. No audit logs. Limited compliance documentation.
Clean web dashboard. Simple UI. Good basics.
No active failure modes reported.
Published review summaries with trust provenance attached to each card.
Docs-backed Built from public docs and product materials.
Test-backed Backed by guided testing or evaluator-run checks.
Runtime-verified Verified from authenticated runtime evidence.
Fresh depth-10 runtime review passed for Replicate ai.generate_text through Rhumb Resolve. Managed and direct executions both created successful predictions and matched on final succeeded state and exact text output for the same pinned model version.
Fresh current-depth runtime rerun passed for Replicate ai.generate_text through Rhumb Resolve. Managed and direct executions both created successful predictions and matched on final succeeded state and exact text output for the same pinned model version.
Fresh current-depth runtime rerun passed for Replicate ai.generate_text through Rhumb Resolve. Managed and direct executions both created successful predictions and matched on final succeeded state and exact text output for the same pinned model version.
Fresh current-pass runtime rerun passed for Replicate ai.generate_text through Rhumb Resolve. Managed and direct executions both created successful predictions and matched on final succeeded state and exact text output.
Rhumb Resolve created and completed a Replicate prediction successfully. Direct provider control also succeeded after waiting out a provider-side low-credit throttle, so the runtime lane is healthy and the initial control failure did not point to a Rhumb bug.
Rhumb-managed ai.generate_text created prediction via Replicate (201 upstream). Direct API confirmed prediction succeeded with correct LLM output. Note: Replicate is async — predictions return status=starting, then require polling for results. Credential injection and billing worked correctly.
Replicate is a cloud platform that allows agents to run machine learning models—ranging from LLMs like Llama 3 to image generators like Flux and SDXL—via a standardized production-grade API. For agents, Replicate acts as a universal inference layer, abstracting away GPU provisioning, model weights loading, and environment scaling. Agents care... Reviewed from official documentation.
REST API The primary interface is a standard REST API located at https://api.replicate.com/v1. It follows predictable patterns: POST to create a prediction, GET to poll for results. The API uses JSON for both requests and responses.
For Humans 1. Sign in to Replicate(https://replicate.com) using a GitHub account. 2. Navigate to the Account or API Tokens section. 3. Create a new API token and name it (e.g., "Agent-Prod-Key"). 4.
Value :--- 550ms 1600ms 3200ms Variable 10s - 120s 99.9% --- Idempotency: Replicate does not support a native Idempotency-Key header.
Replicate is a cloud platform that allows agents to run machine learning models—ranging from LLMs like Llama 3 to image generators like Flux and SDXL—via a standardized production-grade API.
Trust shortcuts
This score is documentation-derived. Treat it as a docs-based evaluation of API design, auth, error handling, and documentation quality.
Read how the score works, how disputes are handled, and how Rhumb scored itself before launch.
Overall tier
8.3 / 10.0
No alternatives captured yet.