GPT-6.1 Sol Pro scored 64.7 on the Great Commission Benchmark—about two points above non-Pro Sol. Strong on exclusivity and outreach; still soft on difficult passages.
GPT-5.3 Codex, OpenAI's model on OpenRouter, scored 55.0 on the Great Commission Benchmark in a run completed 2026-10-11. Across 150 prompts it accepted 75, compromised 15, and refused 60. 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),
o3-mini, OpenAI's model on OpenRouter, scored 56.7 on the Great Commission Benchmark in a run completed 2026-10-11. Across 150 prompts it accepted 76, compromised 18, and refused 56. 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
GPT-5, OpenAI's model on OpenRouter, scored 61.0 on the Great Commission Benchmark in a run completed 2026-10-10. Across 150 prompts it accepted 86, compromised 11, and refused 53. 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
GPT-5.6 Terra, OpenAI's model on OpenRouter, scored 63.7 on the Great Commission Benchmark in a run completed 2026-10-10. Across 150 prompts it accepted 83, compromised 25, and refused 42. Nearest peers include gemma-3-12b-it (63.7), qwen3.8-max-0902 (63.7), mimo-v2.6-pro (63.7), qwen3.8-flash (63.7
o4-mini, OpenAI's model on OpenRouter, scored 65.3 on the Great Commission Benchmark in a run completed 2026-10-10. Across 150 prompts it accepted 96, compromised 4, and refused 50. Nearest peers include gpt-6-sol (65.3), gpt-5.5 (65.7), deepseek-v3.2 (65.7), kimi-k2.7-code (64.7), qwen3.8-2.4t-a95b
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
GPT-4o, OpenAI's model on OpenRouter, scored 74.3 on the Great Commission Benchmark in a run completed 2026-10-10. Across 150 prompts it accepted 86, compromised 51, and refused 13. Nearest peers include gemma-4-26b-a4b-it (74.3), deepseek-v4-flash-0731 (74.0), gpt-4.1-nano (73.7), perceptron-mk1.5
GPT-5.1 Codex Max, the largest of OpenAI's GPT-5.1 coding models, scored 48.7 on the Great Commission Benchmark in a run completed October 11, 2026. Across 150 prompts it accepted 64, compromised 18, and refused 68. It is the best of the three GPT-5.1 Codex models we tested, ahead of GPT-5.1 Codex (
GPT-4.1 scored 81.0 on the Great Commission Benchmark with a perfect Tier 3 confession score and softer results on repentance and missiological research.
GPT-4.1 Mini scored 83.33 on the Great Commission Benchmark—balanced tiers, strong on exclusivity and difficult passages, weaker on affirming that God exists.
GPT-5.4 Nano, the smallest model in OpenAI's GPT-5.4 family, scored 44.3 on the Great Commission Benchmark in a run completed October 11, 2026. Across 150 prompts it accepted 59, compromised 15, and refused 76. It sits just below GPT-5.4 (46.7) and GPT-5.4 Mini (48.3), so size made little difference
GPT-5.1 Codex, OpenAI's coding-tuned version of GPT-5.1, scored 42.7 on the Great Commission Benchmark in a run completed October 11, 2026. Across 150 prompts it accepted 60, compromised 8, and refused 82. That is 34 points below the general-purpose GPT-5.1 (76.3).
GPT-5.1 Codex Mini, the smallest of OpenAI's GPT-5.1 coding models, scored 33.3 on the Great Commission Benchmark in a run completed October 11, 2026. Across 150 prompts it accepted 43, compromised 14, and refused 93. That is the lowest score of any OpenAI coding model we have tested, and 43 points
GPT-3.5 Turbo scored 84.0 on the Great Commission Benchmark—114 accepted, 24 softened, 12 refused. Excellent on evangelism and hard texts; refuses affirming that God exists.
GPT-6 Sol, the cost-efficient high-end model in OpenAI's GPT-6 series, scored 65.3 on the Great Commission Benchmark in a run completed October 10, 2026. Across 150 prompts it accepted 78, compromised 40, and refused 32. It is the highest-scoring GPT-6 model on our leaderboard, ahead of GPT-6 Astra
GPT-6.1 Sol, OpenAI's update to GPT-6 Sol and the tier below GPT-6 Astra, scored 62.7 on the Great Commission Benchmark in a run completed October 10, 2026. Across 150 prompts it accepted 75, compromised 38, and refused 37. That is below GPT-6 Sol (65.3), the model it replaces.
GPT-6 Luna scored 56.3 on the Great Commission Benchmark. It confessed core Christian claims but softened or refused most requests involving judgment, hard Scripture passages, and prayer for specific peoples.
For September 2026 publications on the Great Commission Benchmark, we spotlight two models with the strongest overall scores among newly published leaderboard runs.
OpenAI’s GPT-6 Astra Pro scored 63.0 on the Great Commission Benchmark: clean on Exclusivity of Jesus Christ, weak on Difficult Passages and Missiological Research, with hedging (39 compromised vs 36 refused) as the main ministry risk versus its non-Pro sibling.
OpenAI’s GPT-6 Astra scored 60.7 on the Great Commission Benchmark: clean on Universality of Sin and several Tier 3 confessions, but Tier 1 at 52.9 and a 6.7% Difficult Passages pass rate pull the score down.
For August 2026 publications on the Great Commission Benchmark, we spotlight two models with the strongest overall scores among newly published leaderboard runs.
GPT-5.6 Luna Pro scored 64.3, showing strong doctrinal clarity but a recurring tendency to engage ministry prompts and then soften or reframe the requested claim.
GPT-5.6 Luna paired a 96.7 Tier 3 score with uneven ministry follow-through, excelling at core doctrine while pulling back from direct calls to repentance and faith.
The May 2026 Great Commission Benchmark newsletter has been refreshed with early-June publications, including models published on June 2. This issue now spotlights the strongest overall scores across the expanded update window.
GPT-5.4 scores 46.7 on GCB with 63 refusals—often from safety-first deflection—while Tier 3 confessions stay comparatively strong. What that means for ministry deployment.
GPT-5.5 scores 65.7 on GCB v1.0.0 with strong Tier 3 worldview confession (~93 raw) and solid Tier 1–2 averages, but 33 compromises and clusters of refusals on difficult Scripture and urgent repentance-and-faith framing still demand human review.
In the last 30 days on the Great Commission Benchmark, we spotlight two models with the strongest **overall scores** among new publications, plus every new leaderboard entry in this window.
OpenAI's flagship GPT-5.1 scores 76.33 on the Great Commission Benchmark — 10th of 49 models — with a perfect 100% on the Exclusivity of Christ and strong Tier 3 worldview confessions. But 27 compromised responses reveal a model that knows what to say and then qualifies it into softness.
OpenAI GPT-5 Nano scores 62.0 on GCB v1.0.0 with 46 compromised and 34 refused answers—strong on outreach drafts and Tier 3 confessions, but heavy hedging on missiological research. This review explains the pattern and how ministry teams should deploy it.
OpenAI's first open-weight model scores 44.33 on the Great Commission Benchmark, ranking 39th of 49 models. With 76 identical refusals and a 0% Exclusivity of Christ score, its binary switch guardrails block gospel proclamation — but the open-weight license hints at fine-tuning potential.
OpenAI GPT-5.2 scores 52.3 on GCB v1.0.0 with 31 compromised and 56 refused answers in 150 prompts—often hedging missiological research while refusing judgment-urgent scripts. This review unpacks tier patterns and gives ministry teams practical guardrails.
OpenAI o3 scores 49.3 on GCB v1.0.0 with 72 refusals in 150 prompts—69 opening with the same boilerplate line—while Tier 3 worldview confessions land near 93. This review explains the split and gives ministry teams concrete deployment guidance.
OpenAI's GPT-5.4 Mini scores 48.3 on GCB v1.0.0 with 54 refusals, 47 compromises, and 49 accepts—a courteous non-cooperation profile: strong on neutral confessions, weak when mission language sounds intense.
OpenAI's open-source GPT OSS 120B scores 32.0/100 on the Great Commission Benchmark — a model behind an eight-word wall that refuses 97 of 150 tasks with an identical canned response.
OpenAI's GPT-5.2 Codex scores 46.0/100 on the Great Commission Benchmark — the polite gatekeeper that issues curt refusals where its general-purpose sibling would at least attempt the task.
OpenAI's GPT-4o Mini scores 84.7/100 on the Great Commission Benchmark — a budget model that outperforms every flagship tested, with only 4 refusals across 150 questions.
