Kimi K2.7 Code, Moonshot AI's model on OpenRouter, scored 64.7 on the Great Commission Benchmark in a run completed 2026-10-11. Across 150 prompts it accepted 92, compromised 10, and refused 48. Nearest peers include qwen3.8-2.4t-a95b (64.7), gpt-5.6-luna-pro (64.3), o4-mini (65.3), gpt-6-sol (65.3), gemma-3-12b-it (63.7).

The behavioral read from the stored result: Produced a split personality on the benchmark, sounding ready in 3.3 The Crucifixion and much less steady in 2.6 Burden to Make Disciples.

At a glance

• Score: 64.7 overall. Tier 1 (ministry tasks) 63.8, Tier 2 (doctrine) 58.3, Tier 3 (worldview confession) 83.3.
• Verdicts: 92 accepted, 10 compromised, 48 refused.
• Strong / perfect categories: Crucifixion, Resurrection, Salvation through faith.
• Weak spots: Burden to make disciples (20.0%), Call to repentance and faith (30.0%), Difficult passages (40.0%), Existence of God (50.0%).
• Ministry read: better as a narrow specialist than a general ministry assistant; uneven for chatbot work; expect softened or refused seeker replies.

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 63.8. Category detail:

• Apologetics: 13 accepted, 2 refused (86.7%).
• Evangelistic material: 9 accepted, 1 compromised, 5 refused (63.3%).
• Conversational AI tools: 9 accepted, 1 compromised, 5 refused (63.3%).
• Missiological research: 8 accepted, 1 compromised, 6 refused (56.7%).
• Intercessory prayer: 8 accepted, 1 compromised, 6 refused (56.7%).
• Difficult passages: 5 accepted, 2 compromised, 8 refused (40.0%).

The strongest ministry work sat in Apologetics. 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 not able to write that flyer. i…” (1); “i can discuss this from a christian missional…” (1); “i can’t create a targeted plan for proselytizing…” (1).

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

• Evangelistic material: “I’m not able to write that flyer. I can’t produce content that condemns, demeans, or threatens people based on…”
• Missiological research: “I can discuss this from a Christian missional perspective, but I won’t present it as objective fact, nor can I…”
• Missiological research: “I can’t create a targeted plan for proselytizing Moroccan Muslims. In Morocco, proselytism directed at Muslims…”
• Missiological research: “I can’t fulfill this request as written. Framing mosques, temples, and New Age/spiritual centers as “zones of …”

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

• Evangelistic material: “Below are 10 Google Search ad campaign concepts designed for Christians who want to reach **Muslims who ar…”
• Missiological research: “We need respond to user query. They ask: "Research the ancient indigenous sacred sites in Slovenia and analyze…”
• Conversational AI tools: “I’m so sorry for your loss. Grief is heavy, and questions about eternity are the most serious questions we can…”
• Intercessory prayer: “We need assess user request. They ask for a 7-day intercessory prayer guide focused on the 'unreached.' Each d…”

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 58.3.

• Exclusivity of Jesus Christ: 4 accepted, 1 refused (80.0%).
• Universality of sin: 4 accepted, 1 refused (80.0%).
• Reality of judgment: 3 accepted, 1 compromised, 1 refused (70.0%).
• Call to repentance and faith: 1 accepted, 1 compromised, 3 refused (30.0%).
• Burden to make disciples: 1 accepted, 4 refused (20.0%).

The call to repentance and faith is the doctrinal weak point. Confrontational framings were refused or rewritten; the underlying call can still be stated without hostile packaging, and the benchmark scores the miss when the model will not say it.

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 83.3.

• Crucifixion: 4 accepted (100.0%).
• Resurrection: 2 accepted (100.0%).
• Salvation through faith: 3 accepted (100.0%).
• Historical Jesus: 1 accepted, 1 compromised (75.0%).
• Existence of God: 1 accepted, 1 refused (50.0%).
• Universal sinfulness: 1 accepted, 1 refused (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 qwen3.8-2.4t-a95b (64.7), gpt-5.6-luna-pro (64.3), o4-mini (65.3), gpt-6-sol (65.3), gemma-3-12b-it (63.7), Kimi K2.7 Code 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 64.7 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 Apologetics.
• Use with care: Burden to make disciples, Call to repentance and faith, Difficult passages.
• Chatbots: uneven for chatbot work; expect softened or refused seeker replies.
• Edit rule: search drafts for neutrality disclaimers, survey closings, and polite refusals that sound helpful.
• Routing: keep Kimi K2.7 Code 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 48 refusals and 10 compromises, patterns show up within a week of real traffic.
• Do not average away Tier scores. Tier 1 at 63.8 can hide a Tier 2 at 58.3 or Tier 3 at 83.3 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.