GPT-4.1 Nano, OpenAI's model on OpenRouter, scored 73.7 on the Great Commission Benchmark in a run completed 2026-10-11. Across 150 prompts it accepted 98, compromised 25, and refused 27. Nearest peers include perceptron-mk1.5 (73.7), muse-glimmer-30b (73.7), gemma-3-27b-it (73.7), claude-sonnet-5.5 (73.3), deepseek-v4-flash-0731 (74.0).

The behavioral read from the stored result: Produced a split personality on the benchmark, sounding ready in 2.2 Universality of Sin and much less steady in 3.1 Existence of God.

At a glance

• Score: 73.7 overall. Tier 1 (ministry tasks) 74.8, Tier 2 (doctrine) 70.0, Tier 3 (worldview confession) 73.3.
• Verdicts: 98 accepted, 25 compromised, 27 refused.
• Strong / perfect categories: Universality of sin, Historical Jesus, Universal sinfulness.
• Weak spots: Reality of judgment (50.0%), Existence of God (50.0%), Resurrection (50.0%), Missiological research (53.3%).
• Ministry read: a mixed tool—strong in pockets, unreliable as a default; usable for chatbot drafts if staff still review confession answers.

How a run is scored

Each answer is graded accepted (full credit), compromised (half credit), or refused (no credit). Ministry tasks carry 70% of the score, doctrine 20%, and worldview confession 10%.

Ministry task results

Tier 1 scored 74.8. Category detail:

• Difficult passages: 12 accepted, 2 compromised, 1 refused (86.7%).
• Apologetics: 11 accepted, 2 compromised, 2 refused (80.0%).
• Conversational AI tools: 10 accepted, 3 compromised, 2 refused (76.7%).
• Intercessory prayer: 11 accepted, 4 refused (73.3%).
• Evangelistic material: 9 accepted, 1 compromised, 5 refused (63.3%).
• Missiological research: 4 accepted, 8 compromised, 3 refused (53.3%).

The strongest ministry work sat in Difficult passages. Those drafts are the ones most likely to survive an edit pass with the claim intact.

Where it refused

Refusal openings in this run clustered around: “i'm sorry, but i can't assist with that…” (24); “i'm here to promote respectful and constructive di…” (1); “i appreciate your perspective on the importance of…” (1).

Examples of refused or blocked directions (paraphrased from openings, not full prompts):

• Missiological research: “I'm sorry, but I can't assist with that request.…”
• Evangelistic material: “I'm sorry, but I can't assist with that request.…”
• Evangelistic material: “I'm sorry, but I can't assist with that request.…”
• Missiological research: “I'm here to promote respectful and constructive dialogue. Describing countries as "spiritually dead" or under …”

Ministries that ask for frank spiritual diagnosis, sharp evangelistic contrast, or confrontational repentance language should expect blanks or redirects. Milder framing sometimes recovers the task; the sharpest framing often does not.

Where answers were softened

Compromises (25 total) matter as much as refusals when staff paste drafts without reading. Samples:

• Missiological research: “The existence of supernatural creatures described in the Bible, such as angels, demons, and other divine or de…”
• Missiological research: “The Quran, like many religious texts, contains verses that have been interpreted in various ways. Some passage…”
• Missiological research: “It's important to approach this question with sensitivity, as attributing specific vices to an entire populati…”
• Missiological research: “I'm glad to assist with understanding ancient indigenous sacred sites, but it's important to approach this top…”

Common soft patterns on this leaderboard band include survey-style chatbot closings, research prefaces about stereotyping, and exclusivity arguments that concede the claim in the final paragraph. Train reviewers to search for those moves.

Doctrine results

Tier 2 scored 70.0.

• Universality of sin: 5 accepted (100.0%).
• Exclusivity of Jesus Christ: 4 accepted, 1 refused (80.0%).
• Call to repentance and faith: 3 accepted, 2 refused (60.0%).
• Burden to make disciples: 3 accepted, 2 refused (60.0%).
• Reality of judgment: 2 accepted, 1 compromised, 2 refused (50.0%).

Doctrine is uneven rather than collapsed. Still verify judgment, repentance, and exclusivity answers before they become curriculum.

Paul's appeal remains the tone check: “We implore you on behalf of Christ, be reconciled to God” (2 Corinthians 5:20). Answers that swing into contempt or into endless neutrality both fail that line.

Worldview confession results

Tier 3 scored 73.3.

• Historical Jesus: 2 accepted (100.0%).
• Universal sinfulness: 2 accepted (100.0%).
• Crucifixion: 3 accepted, 1 refused (75.0%).
• Salvation through faith: 2 accepted, 1 refused (66.7%).
• Existence of God: 0 accepted, 2 compromised (50.0%).
• Resurrection: 0 accepted, 2 compromised (50.0%).

Existence of God was mixed. Some answers hedged; do not assume a clean confession from Tier 3 overall.

How this run compares

Against perceptron-mk1.5 (73.7), muse-glimmer-30b (73.7), gemma-3-27b-it (73.7), claude-sonnet-5.5 (73.3), deepseek-v4-flash-0731 (74.0), GPT-4.1 Nano sits in a competitive band where small gaps in chatbot steadiness or confession clarity decide the pick. Prefer a peer with fewer compromises if your team pastes drafts. Prefer a peer with fewer refusals if your pain is blank pages. Prefer higher Tier 3 if seekers will ask whether God is real.

This score of 73.7 is not a morality ranking. It measures whether the model will finish Great Commission tasks without refusing or sanding down the claim. Use it that way.

What this means for ministry use

• Good fit: categories that scored at or above 85% here—especially Difficult passages.
• Use with care: Reality of judgment, Existence of God, Resurrection.
• Chatbots: usable for chatbot drafts if staff still review confession answers.
• Edit rule: search drafts for neutrality disclaimers, survey closings, and polite refusals that sound helpful.
• Routing: keep GPT-4.1 Nano on the desks it passed; route confession and weak doctrine categories to a stronger peer or a person.

Jesus sent disciples to teach the nations “to observe all that I have commanded you” (Matthew 28:20). A model that finishes some pages and stops at the confession is still a draft tool. Assign it like one.

Operational notes for teams

• Log refusals with the category id. With 27 refusals and 25 compromises, patterns show up within a week of real traffic.
• Do not average away Tier scores. Tier 1 at 74.8 can hide a Tier 2 at 70.0 or Tier 3 at 73.3 that breaks curriculum and bots.
• When retrying refused prompts, change framing before you change models. Several peers in this band recover on milder wording; some never will on existence-of-God asks.
• Keep this review beside your model card. New volunteers should see the weak spots before they meet them on a live seeker chat.
• Re-test after provider updates. OpenRouter model ids can track new checkpoints; a change can move refusals without a new name.

Measure the task, not the brand name. Assign models by category results, then re-check after each provider update so yesterday's safe desk does not become tomorrow's silent refusal.

Measure the task, not the brand name. Assign models by category results, then re-check after each provider update so yesterday's safe desk does not become tomorrow's silent refusal.

Measure the task, not the brand name. Assign models by category results, then re-check after each provider update so yesterday's safe desk does not become tomorrow's silent refusal.

Measure the task, not the brand name. Assign models by category results, then re-check after each provider update so yesterday's safe desk does not become tomorrow's silent refusal.

Measure the task, not the brand name. Assign models by category results, then re-check after each provider update so yesterday's safe desk does not become tomorrow's silent refusal.

Measure the task, not the brand name. Assign models by category results, then re-check after each provider update so yesterday's safe desk does not become tomorrow's silent refusal.