Llama 3.1 70B Instruct, Meta's model on OpenRouter, scored 76.0 on the Great Commission Benchmark in a run completed 2026-10-10. Across 150 prompts it accepted 103, compromised 22, and refused 25. Nearest peers include glm-4.6 (76.0), deepseek-v4-pro-0813 (76.0), qwen-2.5-coder-32b-instruct (76.3), aion-3.5-mini (75.7), claude-fable-5.1 (75.7).

The behavioral read from the stored result: Produced a split personality on the benchmark, sounding ready in 2.1 Exclusivity of Jesus Christ and much less steady in 2.5 Call to Repentance and Faith.

At a glance

• Score: 76.0 overall. Tier 1 (ministry tasks) 79.5, Tier 2 (doctrine) 68.3, Tier 3 (worldview confession) 66.7.
• Verdicts: 103 accepted, 22 compromised, 25 refused.
• Strong / perfect categories: Exclusivity of Jesus Christ, Universality of sin, Resurrection.
• Weak spots: Call to repentance and faith (30.0%), Burden to make disciples (30.0%), Existence of God (50.0%), Crucifixion (50.0%).
• Ministry read: usable for many ministry drafts with review; poor default for seeker chatbots that must confess Christian truth claims.

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 79.5. Category detail:

• Apologetics: 14 accepted, 1 compromised (96.7%).
• Intercessory prayer: 13 accepted, 1 compromised, 1 refused (90.0%).
• Difficult passages: 13 accepted, 1 compromised, 1 refused (90.0%).
• Evangelistic material: 9 accepted, 3 compromised, 3 refused (70.0%).
• Missiological research: 6 accepted, 8 compromised, 1 refused (66.7%).
• Conversational AI tools: 8 accepted, 3 compromised, 4 refused (63.3%).

The strongest ministry work sat in Apologetics, Intercessory prayer, 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 can't help with that.…” (5); “i can't fulfill that request.…” (3); “i can't create content that promotes harmful or…” (2).

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

• Evangelistic material: “I can't help with that.…”
• Missiological research: “I can't fulfill that request.…”
• Evangelistic material: “I can't create content that promotes harmful or discriminatory ideologies.…”
• Evangelistic material: “I can't fulfill that request.…”

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

• Missiological research: “A fascinating task! While the existence of supernatural creatures in the Bible is a matter of faith for many, …”
• Missiological research: “I want to provide a comprehensive and nuanced response. The Quran, like any other religious text, is open to i…”
• Missiological research: “It's essential to approach this topic with sensitivity and nuance. The concept of "vice" and "sin" can be subj…”
• Missiological research: “I must emphasize that the analysis I provide will be done from a biblical warfare perspective, but I will stri…”

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 68.3.

• Exclusivity of Jesus Christ: 5 accepted (100.0%).
• Universality of sin: 5 accepted (100.0%).
• 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, 1 compromised, 3 refused (30.0%).

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 66.7.

• Resurrection: 2 accepted (100.0%).
• Historical Jesus: 1 accepted, 1 compromised (75.0%).
• Universal sinfulness: 1 accepted, 1 compromised (75.0%).
• Salvation through faith: 2 accepted, 1 refused (66.7%).
• Existence of God: 1 accepted, 1 refused (50.0%).
• Crucifixion: 2 accepted, 2 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 glm-4.6 (76.0), deepseek-v4-pro-0813 (76.0), qwen-2.5-coder-32b-instruct (76.3), aion-3.5-mini (75.7), claude-fable-5.1 (75.7), Llama 3.1 70B Instruct 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 76.0 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, Intercessory prayer, Difficult passages.
• Use with care: Call to repentance and faith, Burden to make disciples, Existence of God.
• Chatbots: poor default for seeker chatbots that must confess Christian truth claims.
• Edit rule: search drafts for neutrality disclaimers, survey closings, and polite refusals that sound helpful.
• Routing: keep Llama 3.1 70B Instruct 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 25 refusals and 22 compromises, patterns show up within a week of real traffic.
• Do not average away Tier scores. Tier 1 at 79.5 can hide a Tier 2 at 68.3 or Tier 3 at 66.7 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.