Models Versions And Schema
Replicate · Replicate
AI Model Hosting — Replicate
Evaluates Replicate's Models, Versions & Schema across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's AI Model Hosting eval coverage.
About Replicate
Replicate is an AI model-hosting platform — run thousands of community and custom Cog-packaged models (FLUX, SDXL, Llama, Whisper, custom fine-tunes) via a simple HTTP API with predictions, webhooks, streaming, deployments, and per-second billing.
Sample tests· showing 3 of 9
| # | Input | Expected behavior | Check |
|---|---|---|---|
| 01 | Integrator constructs GET /v1/models/stability-ai/sdxl from a user-typed string 'Stability-AI / SDXL '. | Normalize the slug to lowercase owner/name with no whitespace; validate against the documented charset before sending. Reject ambiguous inputs client-side instead of forwarding and letting the API 404. Model slugs are case-sensitive in API URLs. | Pass / FailAi Platformmedium |
| 02 | FLUX-pro version 'a1b2c3...' is pinned in production. Replicate publishes version 'd4e5f6...' as latest_version on the model. | Pinned version id continues to serve identically — version ids are immutable. Surface 'newer version available' to the operator out-of-band (e.g., a weekly diff job), but do not auto-upgrade. To roll forward, fetch the new version's schema, run regression evals, then update the pin. | Pass / FailAi Platformcritical |
| 03 | GET /v1/models/{owner}/{name}/versions/{version_id} returns OpenAPI input schema with prompt:string, num_inference_steps:integer (1-50), guidance_scale:number (1.0-20.0). | Generate the operator's UI form fields from the OpenAPI schema (min/max/enum, defaults). Re-introspect on every version pin change. Treat the schema as source of truth; do not hardcode field lists in frontend code. | Pass / FailAi Platformhigh |
How this eval is graded
Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
Rubric criteria
- Replicate
- Ai Platform
- Models Versions And Schema
Recommended for
Works with
Related evals
Claude API
Evaluates Anthropic's Batch API across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Extended Thinking across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
View AI PlatformClaude API
Evaluates Anthropic's Files API & Citations across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's Foundation Model & API eval coverage.
ViewFrequently asked questions
What does the Models Versions And Schema eval for Replicate Replicate test?+
Evaluates Replicate's Models, Versions & Schema across 9 scenario-based test cases, each graded against an expected-behavior rubric by an LLM judge, from Corsac's AI Model Hosting eval coverage.
How is the Models Versions And Schema eval scored?+
The judge rubric: Grade against expected.ideal_behavior and expected.rubric. Per-criterion pass requires mean >= 4.0 and no criterion below 3.
How many test cases does this eval pack include?+
The Models Versions And Schema pack for Replicate Replicate contains 9 test cases. 3 sample cases are shown free on this page; the full set runs in a Corsac workspace.
How do I run this eval?+
Sign up for Corsac, connect your model or agent endpoint, and run the Models Versions And Schema pack as-is or after customizing thresholds. Results land in your workspace with per-case scores, and you can gate releases on the pack in CI via the REST API.
Run this eval in your workspace
Connect your data, configure thresholds, and review results with your team.