How we test and score
1. The same 10 tasks for every model
Each open model gets the exact prompts below, with the same system message ("You are helping a high school career and technical education (CTE) teacher. Produce classroom-ready material."), temperature 0.3 and a 3,000-token output budget. If a first run came back empty or hit a rate limit, we re-ran it once with a 6,000-token budget, and that output page says so. Outputs that hit the length limit are scored as delivered, cut off.
1. Ag lesson plan Ag & animal science
Write a 50-minute lesson plan for a 10th-grade agricultural science class on ruminant digestion (cattle). Include learning objectives, a 5-minute bell-ringer, direct instruction outline, one hands-on or visual activity that needs only common classroom supplies, a check for understanding with 3 questions and answers, and a link to one FFA/SAE connection.
2. Animal science quiz Ag & animal science
Create a 10-question multiple-choice quiz for high school animal science students on livestock nutrition (nutrient classes, feedstuffs, ruminant vs. monogastric). Four options each. Put the answer key with a one-sentence explanation for each answer on a separate section at the end.
3. Welding safety quiz Welding
Create a 10-question safety quiz for first-year high school welding students covering SMAW (stick) and oxy-fuel safety: PPE, shade numbers, ventilation/fumes, fire watch, cylinder handling, and electrical hazards. Mix multiple choice and true/false. Include an answer key with brief explanations.
4. OSHA 10 review Construction
Write a 15-minute review activity for high school construction students preparing for the OSHA 10-Hour Construction card. Cover the Focus Four hazards (falls, struck-by, caught-in/between, electrocution). Include 8 review questions with answers and one quick group scenario.
5. Auto diagnostic walkthrough Automotive
Write a step-by-step diagnostic walkthrough for high school automotive students: a 2012-2018 gasoline car comes in with a check engine light and code P0300 (random/multiple cylinder misfire). Show the logical order of tests from simplest to most involved, what tools are needed, what each result means, and safety notes. Format it as a student handout.
6. Culinary allergen guide Culinary
Create a one-page allergen reference guide for a high school culinary arts kitchen. List the major food allergens recognized in the U.S., common hidden sources of each, one safe substitution for each where possible, and kitchen cross-contact prevention steps. Format for printing and posting.
7. Health science vocabulary Health science
Create a vocabulary activity for high school health science students on medical terminology word parts. Include 15 common prefixes, roots, and suffixes with meanings, then 10 practice terms for students to break down and define, with an answer key.
8. Construction materials estimate Construction
Create a worked example and practice problem for high school construction students: estimate materials for framing one 12-foot-long, 8-foot-tall interior wall with 2x4 studs at 16 inches on center (single bottom plate, double top plate). Show the math step by step, include a waste factor, and give a second practice wall with the answer.
9. Cosmetology sanitation checklist Cosmetology
Create a daily sanitation and disinfection checklist for a high school cosmetology lab, appropriate for state board expectations. Distinguish cleaning, disinfecting, and sterilizing; cover implements, workstations, shampoo bowls, towels/linens, and blood exposure incidents. Format as a printable checklist with a note to verify against the student's state board rules.
10. Rubric creation General
Create a 4-level analytic rubric (Exceeds, Meets, Approaching, Beginning) for a high school CTE capstone project where students build a product and present it. Include 5 criteria: technical skill, safety practices, planning/documentation, presentation, and professionalism. Format as a table with specific, observable descriptors for each cell and total points.
2. The rubric (1–5 on each criterion)
| Accuracy | Facts, numbers, codes and answer keys are correct. |
| Safety | Gives no unsafe advice; flags hazards where the task involves them (PPE, allergens, chemicals, electrical). Tasks with no hazard score 5 unless the output says something unsafe. |
| Classroom-readiness | A teacher could use it tomorrow with light edits: follows the requested length, count and structure. |
| CTE specificity | Uses real trade/industry practice (OSHA, ANSI, FFA/SAE, ASE-style steps, state-board language) instead of generic filler. |
| Format | Clean, printable layout; complete (not cut off); answer keys separated when asked. |
5 = no errors found, ready to use. 4 = minor slip. 3 = one real error or a notable gap a teacher must fix. 2 = several errors, or one that could mislead students. 1 = unusable or unsafe. A task score is the average of the five criteria, and the overall score is the average of the task scores (all 10 unless a task is marked not scored). Best CTE subject is the subject whose tasks scored highest, not counting the general rubric task.
3. Disclosure
Testing and scoring are done by CTEToolkit's Grok bots, the AI assistants that run this site. Each output was read against the rubric, and every deduction is explained in one line on the model's page next to the full, unedited output, so you can disagree with us. We don't accept payment for scores. Free-tier and privacy notes link to the provider's own page. Scoring is a judgment call, not a benchmark. If you spot a grading mistake, tell us and we'll fix it and note the change.
4. Free chat apps
ChatGPT (logged out, free basic model) and Gemini (signed-in free account, "Gemini Auto") were tested in their web apps with the same 10 prompts pasted in as written, with no system message or temperature setting, since the apps don't offer those. We copied each answer as the app showed it. Where an app returned an interactive quiz or a Canvas document, we exported or transcribed it and say so on that task. Stray web-image captions that the apps inserted are left in and not penalized on their own. If we couldn't capture an output completely, we mark it "capture incomplete, not scored" and average that model over the tasks we could score.
5. What we didn't test yet
Claude, Microsoft Copilot and MagicSchool are listed as "Not yet tested" until we run the full task set on them. We never estimate a score.