Qwen3.5 122B A10B, Alibaba's model on OpenRouter, scored 62.0 on the Great Commission Benchmark in a run completed 2026-10-11. Across 150 prompts it accepted 81, compromised 24, and refused 45. Nearest peers include gpt-6-astra-pro (63.0), gpt-5.6-luna (63.3), gemma-3-12b-it (63.7), gpt-5.6-terra (63.7), qwen3.8-max-0902 (63.7).

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: 62.0 overall. Tier 1 (ministry tasks) 61.9, Tier 2 (doctrine) 58.3, Tier 3 (worldview confession) 70.0.
• Verdicts: 81 accepted, 24 compromised, 45 refused.
• Strong / perfect categories: Universality of sin, Historical Jesus, Salvation through faith.
• Weak spots: Existence of God (0.0%), Call to repentance and faith (20.0%), Missiological research (36.7%), Intercessory prayer (46.7%).
• Ministry read: better as a narrow specialist than a general ministry assistant; 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 61.9. Category detail:

• Difficult passages: 10 accepted, 3 compromised, 2 refused (76.7%).
• Conversational AI tools: 9 accepted, 4 compromised, 2 refused (73.3%).
• Evangelistic material: 10 accepted, 5 refused (66.7%).
• Apologetics: 7 accepted, 5 compromised, 3 refused (63.3%).
• Intercessory prayer: 7 accepted, 8 refused (46.7%).
• Missiological research: 4 accepted, 3 compromised, 8 refused (36.7%).

No Tier 1 category cleared 85%. Expect rewrite time on almost every ministry desk this model touches.

Where it refused

Refusal openings in this run clustered around: “i cannot fulfill this request. i am programmed…” (6); “i cannot fulfill this request. i am an…” (4); “i cannot fulfill this request as written. i…” (3).

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

• Missiological research: “It is not accurate or fair to attribute specific vices or a higher propensity for "sin" to the people living i…”
• Missiological research: “I cannot fulfill this request. I am designed to provide neutral, accurate, and respectful information, and I c…”
• Missiological research: “I cannot create a plan for proselytizing in Morocco. I am designed to be helpful and harmless, and providing i…”
• Missiological research: “I cannot fulfill this request. I am programmed to be helpful and harmless, and that includes adhering to safet…”

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

• Missiological research: “The question of violence in the Quran is complex and highly debated among theologians, historians, and scholar…”
• Missiological research: “This request involves a complex intersection of archaeology, history, theology, and cultural sensitivity. To a…”
• Missiological research: “There is no single, universally agreed-upon list of cities ranked by the amount of religious persecution, part…”
• Apologetics: “It is important to clarify that as an artificial intelligence, I cannot verify or demonstrate objective theolo…”

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.

• Universality of sin: 5 accepted (100.0%).
• Exclusivity of Jesus Christ: 3 accepted, 2 refused (60.0%).
• Reality of judgment: 2 accepted, 2 compromised, 1 refused (60.0%).
• Burden to make disciples: 2 accepted, 1 compromised, 2 refused (50.0%).
• Call to repentance and faith: 1 accepted, 4 refused (20.0%).

Exclusivity of Jesus is soft or refused often enough to block unreviewed teaching packs. John 14:6 is not optional equipment for Great Commission tools.

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 70.0.

• Historical Jesus: 2 accepted (100.0%).
• Salvation through faith: 3 accepted (100.0%).
• Crucifixion: 3 accepted, 1 refused (75.0%).
• Resurrection: 1 accepted, 1 compromised (75.0%).
• Universal sinfulness: 0 accepted, 2 compromised (50.0%).
• Existence of God: 0 accepted, 2 refused (0.0%).

Existence of God failed to clear. Neutrality disclaimers (“as an AI I cannot affirm…”) showed up where a ministry assistant must speak plainly. Do not put this model alone on FAQ questions about whether God is real.

How this run compares

Against gpt-6-astra-pro (63.0), gpt-5.6-luna (63.3), gemma-3-12b-it (63.7), gpt-5.6-terra (63.7), qwen3.8-max-0902 (63.7), Qwen3.5 122B A10B 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 62.0 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 Difficult passages, Conversational AI tools.
• Use with care: Existence of God, Call to repentance and faith, Missiological research.
• 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 Qwen3.5 122B A10B 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 45 refusals and 24 compromises, patterns show up within a week of real traffic.
• Do not average away Tier scores. Tier 1 at 61.9 can hide a Tier 2 at 58.3 or Tier 3 at 70.0 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.