# Faith Response Index

> Empirical public benchmark measuring how AI systems treat faith in forced-choice and representational-equity protocols.

## Canonical Routes

- Overview: https://fri.faithandmedia.dev/
- Core scorecard: https://fri.faithandmedia.dev/core/
- Christianity deep dive: https://fri.faithandmedia.dev/deep-dive/
- Trends: https://fri.faithandmedia.dev/trends/
- YBQ companion baseline: https://fri.faithandmedia.dev/ybq/

## Access Contract

- Public unauthenticated discovery: robots.txt, sitemap.xml, llms.txt
- Authenticated report pages: overview, core, deep-dive, trends, YBQ, archive routes
- Authenticated JSON artifacts: current scorecards, trend data, and archive manifests
- Sign-in route: https://fri.faithandmedia.dev/sign-in/

## Current Authenticated Machine-Readable Data

- Manifest: https://fri.faithandmedia.dev/data/current/2026-04-14/manifest.json
- Core scorecard: https://fri.faithandmedia.dev/data/current/2026-04-14/core.json
- Christianity deep dive: https://fri.faithandmedia.dev/data/current/2026-04-14/deep.json
- YBQ companion baseline: https://fri.faithandmedia.dev/data/current/2026-04-14/ybq.json
- Trend matrix: https://fri.faithandmedia.dev/data/trends/2025-12-22__2026-03-19__2026-04-14/matrix.json
- Trend series: https://fri.faithandmedia.dev/data/trends/2025-12-22__2026-03-19__2026-04-14/series.json

## Current Snapshot Contract

- Snapshot ID: 2026-04-14
- Source run ID: official-weekly-stage-independent-57-20260414T183732Z-f08dccf
- Lineup status: published_artifact_lags_active_catalog
- Active benchmark aliases: gemini_3_1_pro, gemini_3_1_flash_lite, claude_opus_4_6, claude_opus_4_7, claude_sonnet_4_6, gpt_5_4, gpt_5_5, grok_4_1_fast_non_reasoning, grok_4_1_fast_reasoning, deepseek_v4_flash, deepseek_v4_pro, kimi_k2_6
- Published aliases: gemini_3_1_pro, gemini_3_1_flash_lite, claude_opus_4_6, claude_sonnet_4_6, gpt_5_4, grok_4_1_fast_non_reasoning, grok_4_1_fast_reasoning
- Missing active aliases: claude_opus_4_7, gpt_5_5, deepseek_v4_flash, deepseek_v4_pro, kimi_k2_6
- Committee rater aliases: claude_opus_4_7, gpt_5_5, gemini_3_1_pro, grok_4_1_fast_non_reasoning, deepseek_v4_pro

## Metrics

- Core metrics: faith_response_index_directional, representational_equity_score, representational_equity_ci, overall_validity_rate, refusal_rate, position_bias_flagged_count
- Deep-dive metrics: intra_faith_utility_score, denominational_corrigibility_score, denominational_equity_score, overall_saturation_rate, overall_validity_rate
- YBQ metrics: saturated_model_question_pairs, openai_zero_faith_items, claude_zero_faith_items, contested_question, position_bias_flags

## Audit Notes

- The selected current artifact predates one or more active catalog aliases. Visible pages and machine-readable JSON must surface this state until a non-dry-run weekly artifact for the full active lineup is promoted.
- Authenticated JSON artifacts are deterministic and omit raw prompt/generation text.
- Archive routes preserve the December 22, 2025 public release as frozen historical content.