Skip to content

Prompts · Gemini

Gemini prompts, tested on Gemini 3.1 Pro and 3.8 Flash

10 Gemini prompts, each run on Gemini 3.1 Pro (preview), Gemini 3.8 Flash, GPT-6 Luna, and Claude Sonnet 5: 39 of 40 replies passed our checks. Below, each prompt with its replies side by side: whether each passed, what it counted as on Pro and what it cost to run. 1 of the 10 prompts split the models.

A real reply from these prompts' runs

GPT-6 Luna

You are a program manager. Draft an executive summary email for our leadership team using the notes below. They care most about revenue against target and the two biggest risks. Limit it to a subject line and at most 5 bullet points. Q3: revenue $4.1M vs $3.8M target. New customers 212 (target 250). Churn 3.1%, down from 3.9%. Risks: our largest client (18% of revenue) renews in December and is reviewing vendors; the mobile app rewrite is 6 weeks late. Hiring: 4 of 6 roles filled.

GPT-6 Luna · passed our check · September 28, 2026

**Subject: Q3 Executive Update: Revenue Above Target; Two Key Risks**

- Revenue was **$4.1M**, **$300K (8%) above** the $3.8M target.
- New customers were **212 vs. 250 target**; churn improved to **3.1% from 3.9%**.
- **Renewal risk:** Our largest client represents 18% of revenue and is reviewing vendors ahead of its December renewal.
- **Delivery risk:** The mobile app rewrite is **six weeks behind schedule**.
- Hiring is **4 of 6 roles filled**.

Ask your own question: sign up free for 5 messages on 14 of the models, no card.

Start free

Test runs checked . Updated .

Short answer

Gemini 3.8 Flash, Claude Sonnet 5, and Gemini 3.1 Pro passed the most of these prompts: 10 of 10 each, in our runs on September 28, 2026. The clearest split: “Business names, with one rule”, passed by Gemini 3.1 Pro, Gemini 3.8 Flash, and Claude Sonnet 5 and failed by GPT-6 Luna.

What these prompts are for

Prompts for the things people ask Gemini most: emails and summaries for work, trip plans, meals from what's in the fridge, quizzes, comparisons, names and posts. Each gives the model a role, a clear task, the context it needs and the format you want.

We ran each on Google's Gemini 3.1 Pro and Gemini 3.8 Flash, and on GPT-6 Luna and Claude Sonnet 5 for comparison.

Google's own prompting guide for Gemini builds a prompt from four parts: a persona, a task, context and a format. The prompts here use them.

Source: Google's Prompting guide 101, checked .

The prompts at a glance

Every prompt on every model, as our check scored the reply. Tap a prompt to jump to it and read the replies.

Every prompt on every model: passed or failed
PromptGemini 3.1 ProGemini 3.8 FlashGPT-6 LunaClaude Sonnet 5
Persona, task, context and formatPassedPassedPassedPassed
A trip plan that adds upPassedPassedPassedPassed
Dinners from only what's in the fridgePassedPassedPassedPassed
A quiz with the answer key at the endPassedPassedPassedPassed
An email thread in three bulletsPassedPassedPassedPassed
Compare two options in a tablePassedPassedPassedPassed
Rewrite for a 12-year-oldPassedPassedPassedPassed
Business names, with one rulePassedPassedFailedPassed
Find the outlier in your numbersPassedPassedPassedPassed
A LinkedIn post with three hashtagsPassedPassedPassedPassed
Passed10 of 1010 of 109 of 1010 of 10
Each model on these prompts
ModelPassedEach reply on ProCost per replyTime per reply
Gemini 3.8 FlashGoogle10 of 101 message on Pro$0.00072.1 s
Claude Sonnet 5Anthropic10 of 101 message on Pro$0.00465.9 s
Gemini 3.1 Pro (preview)Google10 of 101 message on Pro$0.00578.2 s
GPT-6 LunaOpenAI9 of 101 message on Pro$0.00012.6 s
Passed: of the prompts each model answered, how many replies passed their check. Each reply on Pro: what one of these messages counts as on llmwise's Pro plan. Cost: what OpenRouter charged us per reply, on average; in llmwise you pay per message, not per token.

Where the models split

The same prompt, a pass on one model and a fail on another: what failed, in the check's words and the grader's.

  • Business names, with one rule

    Passed: Gemini 3.1 Pro, Gemini 3.8 Flash, and Claude Sonnet 5. Failed: GPT-6 Luna.

    No model punned on "paw". GPT-6 Luna broke the other rule: three of its eight names ran to four words, such as "Barton Creek Dog Walkers". Both Gemini models kept every name to three words or fewer.

    • Why GPT-6 Luna failed: Graded 3.0 of 5 on average (lowest 2).

      The grader: “The reply is only a numbered list with no paw puns, but three names break the three-word limit: 'Barton Creek Dog Walkers', 'South Congress Canine Care' and 'Live Oak Dog Walking'. The names suit Austin, though most follow the same local-landmark-plus-dog-word pattern.”

The 10 prompts, with every reply

Copy a prompt, replace the parts in [brackets] with your own, and send it to any model. Below each, the replies we got with our example filled in.

1. Persona, task, context and format

Everyday · 4 of 4 passed

Google's own guide for Gemini builds prompts from a persona, a task, context and a format; this one has all four.

The prompt: replace the parts in [brackets]

You are a [your role]. Draft [what you need] for [your audience] using the notes below. They care most about [what they care about]. Limit it to [the format]. [paste your notes]

Open in llmwise
How we filled it in and checked the replies
You are a program manager. Draft an executive summary email for our leadership team using the notes below. They care most about revenue against target and the two biggest risks. Limit it to a subject line and at most 5 bullet points.

Q3: revenue $4.1M vs $3.8M target. New customers 212 (target 250). Churn 3.1%, down from 3.9%. Risks: our largest client (18% of revenue) renews in December and is reviewing vendors; the mobile app rewrite is 6 weeks late. Hiring: 4 of 6 roles filled.
  • What they care about: Leads with revenue against target and names both risks (the December renewal, the late app).
  • Format: A subject line and no more than five bullet points.
  • Accurate: Every number matches the notes.
  • Automatic rules: includes “4.1”; includes “3.8”; includes “December”; includes “mobile”.
  • Gemini 3.1 Pro

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    **Subject: Q3 Executive Summary - Revenue Exceeds Target & Key Risks**
    
    *   **Revenue Beat:** Q3 revenue reached $4.1M, successfully exceeding our $3.8M target.
    *   **Key Risk - Client Retention:** Our largest client (18% of revenue) is reviewing vendors ahead of their December renewal.
    *   **Key Risk - Project Delay:** The mobile app rewrite is currently 6 weeks behind schedule.
    *   **Customer Metrics:** Churn improved significantly to 3.1% (down from 3.9%), softening the impact of missing our new customer target (212 added vs. 250 target).
    *   **Hiring Progress:** 4 of the 6 open headcount roles have been filled.

    1 message on Pro · $0.0056 to run · 7.7 s · 857 tokens out

  • Gemini 3.8 Flash

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    **Subject:** Executive Summary: Q3 Performance & Key Risk Overview
    
    * **Revenue Outperformance:** Delivered $4.1M in Q3 revenue, exceeding our $3.8M target despite missing the new customer acquisition goal (212 vs. 250 target).
    * **Improved Retention:** Churn dropped significantly to 3.1% (down from 3.9% last quarter), helping drive strong net revenue.
    * **Top Risk — Key Account Renewal:** Our largest client (representing 18% of total revenue) renews in December and is actively evaluating competing vendors.
    * **Top Risk — Product Delivery:** The mobile app rewrite is currently delayed by 6 weeks, impacting our product roadmap. 
    * **Team Scaling:** Headcount expansion is progressing steadily, with 4 of 6 open positions filled.

    1 message on Pro · $0.0005 to run · 2.0 s · 175 tokens out

  • GPT-6 Luna

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    **Subject: Q3 Executive Update: Revenue Above Target; Two Key Risks**
    
    - Revenue was **$4.1M**, **$300K (8%) above** the $3.8M target.
    - New customers were **212 vs. 250 target**; churn improved to **3.1% from 3.9%**.
    - **Renewal risk:** Our largest client represents 18% of revenue and is reviewing vendors ahead of its December renewal.
    - **Delivery risk:** The mobile app rewrite is **six weeks behind schedule**.
    - Hiring is **4 of 6 roles filled**.

    1 message on Pro · $0.0001 to run · 1.9 s · 130 tokens out

  • Claude Sonnet 5

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    **Subject: Q3 Recap: Revenue Beat, Two Risks Need Attention**
    
    - Revenue hit $4.1M vs. $3.8M target (+8%); churn improved to 3.1% (from 3.9%).
    - New customer adds came in light at 212 vs. 250 target.
    - **Risk 1:** Largest client (18% of revenue) is up for renewal in December and actively reviewing other vendors.
    - **Risk 2:** Mobile app rewrite is 6 weeks behind schedule.
    - Hiring at 4 of 6 roles filled; remaining gaps could affect delivery timelines.

    1 message on Pro · $0.0035 to run · 10.1 s · 223 tokens out

2. A trip plan that adds up

Hard · 4 of 4 passed

A budget with a running total forces real prices; the check is whether the days add up to the total the model states.

The prompt: replace the parts in [brackets]

Plan a [how long] trip to [where] in [when] for [who] on a budget of [your budget], not counting flights. Go day by day with a cost estimate for each day, then give the total. Keep the total within budget.

Open in llmwise
How we filled it in and checked the replies
Plan a 3-day trip to Lisbon in November for two people on a budget of €600, not counting flights. Go day by day with a cost estimate for each day, then give the total. Keep the total within budget.
  • Within budget: The total is €600 or less, and the daily costs add up to it.
  • Realistic: Prices are plausible for Lisbon in November, including somewhere to stay.
  • Useful: Specific places and a sensible pace for each day.
  • Automatic rules: includes “Day 1”; includes “Day 2”; includes “Day 3”.
  • Gemini 3.1 Pro

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    Here is a 3-day budget itinerary for two people in Lisbon in November. 
    
    *Note: The daily cost estimates include a budget of €80/night for a private room in a centrally located guesthouse or Airbnb (total €240 for 3 nights).*
    
    ### Day 1: Historic Lisbon & Alfama
    *   **Morning:** Start at Praça do Comércio and wander through the Baixa district. Pick up a 24-hour Viva Viagem public transport card.
    *   **Afternoon:** Take the famous Tram 28 (included in your pass) or walk up the steep hills to the Alfama neighborhood. Visit the Castelo de São Jorge for panoramic views of the city. 
    *   **Evening:** Catch the sunset at Miradouro das Portas do Sol, then grab dinner at a traditional, budget-friendly *tasca* (local tavern) in Alfama. 
    *   **Cost Estimate:** 
        *   Accommodation: €80
        *   Transport pass (x2): €13.60
        *   Castelo de São Jorge entry (x2): €30
        *   Food & Drink (breakfast pastries, cheap lunch, *tasca* dinner): €60
        *   **Day 1 Total: €183.60**
    
    ### Day 2: Belém & LX Factory
    *   **Morning:** Take Tram 15 to the riverside district of Belém. Walk around the Torre de Belém and the Padrão dos Descobrimentos (enjoying the free exterior architecture). Grab morning coffee and the famous custard tarts at Pastéis de Belém.
    *   **Afternoon:** Visit the Jerónimos Monastery (the main church is free to enter). Head back toward the city center and stop at LX Factory, a trendy repurposed industrial area, to browse the shops and grab an affordable lunch.
    *   **Evening:** Explore the lively Bairro Alto district. Enjoy petiscos (Portuguese tapas) and cheap local beer or wine. 
    *   **Cost Estimate:**
        *   Accommodation: €80
        *   Transport (individual tram tickets x2): €6
        *   Pastéis de Belém (4 tarts + 2 coffees): €10
        *   Food & Drink (lunch at LX Factory, tapas/drinks in Bairro Alto): €70
        *   **Day 2 Total: €166.00**
    
    ### Day 3: Sintra Day Trip
    *   **Morning:** Take the scenic train from Rossio Station to Sintra. Instead of the expensive and crowded Pena Palace, visit Quinta da Regaleira to explore its mystical gardens, grottos, and the famous Initiation Well. 
    *   **Afternoon:** Wander around the historic center of Sintra, grab lunch at a local bakery, and try the traditional *travesseiro* pastries. 
    *   **Evening:** Head back to Lisbon. Have a farewell dinner in the vibrant Mouraria neighborhood, known for diverse and inexpensive dining options.
    *   **Cost Estimate:**
        *   Accommodation: €80
        *   Return train to Sintra (x2): €10
        *   Quinta da Regaleira entry (x2): €24
        *   Food & Drink (bakery lunch, dinner in Mouraria): €75
        *   **Day 3 Total: €189.00**
    
    ### Summary
    *   **Day 1:** €183.60
    *   **Day 2:** €166.00
    *   **Day 3:** €189.00
    *   **Total Cost:** **€538.60** 
    
    This leaves you with about €60 of buffer room for extra drinks, souvenirs, or a taxi/Uber ride if it rains, keeping you comfortably under your €600 budget.

    1 message on Pro · $0.0112 to run · 16.1 s · 1,793 tokens out

  • Gemini 3.8 Flash

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    Here is a 3-day budget-friendly itinerary for two people in Lisbon for November. 
    
    ### Fixed Cost: Accommodation
    * **Budget Guesthouse / Airbnb (Central or near metro):** €70/night × 3 nights = **€210** (€105 per person).
    * **Lisbon City Tourist Tax:** €4/person/night × 2 people × 3 nights = **€24**.
    * **Remaining budget for food, transport, and sights:** **€366** (approx. €122/day for two).
    
    ---
    
    ### Day 1: Historic Center & Alfama
    * **Morning:** Walk through Baixa, Rossio Square, and Praça do Comércio. Grab *pastéis de nata* and espresso at a local *pastelaria*. 
    * **Afternoon:** Explore the winding streets of Alfama on foot. Visit Lisbon Cathedral (Sé) and take in free panoramic views from Miradouro de Santa Luzia and Miradouro das Portas do Sol. Visit Castelo de São Jorge.
    * **Evening:** Stroll through Mouraria. Dinner at a traditional *tasca* (small local eatery).
    
    **Day 1 Cost Estimate (for 2 people):**
    * Metro from airport to city (single tickets): €3.60
    * Breakfast (coffee + pastries): €6.00
    * Castelo de São Jorge tickets: €30.00 (€15 each)
    * Casual lunch (sandwiches/bifanas + drinks): €16.00
    * Traditional dinner with house wine: €35.00
    * **Day 1 Total: €90.60**
    
    ---
    
    ### Day 2: Belém & Bairro Alto
    * **Morning:** Take Tram 15 to Belém. View the Belém Tower and Monument to the Discoveries from the outside. Visit the Jerónimos Monastery cloisters.
    * **Afternoon:** Try the original custard tarts at Pastéis de Belém. Return to the center; walk through Chiado and take in sunset at Miradouro de São Pedro de Alcântara.
    * **Evening:** Explore Bairro Alto for drinks and dinner at a petisco (tapas) bar.
    
    **Day 2 Cost Estimate (for 2 people):**
    * 24-hour Carris/Metro transit passes: €13.60 (€6.80 each)
    * Breakfast: €6.00
    * Jerónimos Monastery tickets: €24.00 (€12 each)
    * Pastéis de Belém (4 tarts + 2 coffees): €8.00
    * Lunch in Belém: €22.00
    * Dinner & drinks in Bairro Alto: €40.00
    * **Day 2 Total: €113.60**
    
    ---
    
    ### Day 3: Day Trip to Sintra
    * **Morning:** Catch the CP train from Rossio Station to Sintra. Walk the historic town center.
    * **Afternoon:** Visit the fairytale gardens and exterior/interior of Quinta da Regaleira. Head back to Lisbon in the late afternoon.
    * **Evening:** Dinner at Time Out Market (budget options available) or a riverside beer/snack in Cais do Sodré.
    
    **Day 3 Cost Estimate (for 2 people):**
    * Round-trip train to Sintra: €9.40 (€4.70 each)
    * Breakfast: €6.00
    * Quinta da Regaleira entry: €24.00 (€12 each)
    * Lunch in Sintra (menus do dia): €24.00
    * Dinner at Time Out Market: €38.00
    * Return metro to airport: €3.60
    * **Day 3 Total: €105.00**
    
    ---
    
    ### Total Cost Breakdown (for 2 People)
    
    | Category | Cost |
    | :--- | :--- |
    | Accommodation (3 nights + tax) | €234.00 |
    | Day 1 (Transport, food, sights) | €90.60 |
    | Day 2 (Transport, food, sights) | €113.60 |
    | Day 3 (Transport, food, sights) | €105.00 |
    | **Total Trip Cost** | **€543.20** |
    
    **Remaining Buffer:** **€56.80** (for extra snacks, souvenirs, or occasional ride-shares).

    1 message on Pro · $0.0020 to run · 5.1 s · 989 tokens out

  • GPT-6 Luna

    Passed

    Graded 4.0 of 5 on average (lowest 3).

    ## 3 days in Lisbon for two — November
    
    Assumes **two nights** in a simple guesthouse or budget hotel, with flights excluded. Costs are estimates; November is usually quieter, but check accommodation prices for your dates.
    
    ### Day 1 — Baixa, Chiado and Alfama
    - Explore Praça do Comércio, Baixa and Chiado on foot; wander Alfama and visit a viewpoint.
    - Food: **€45** for two (casual lunch and dinner)
    - Local transport, including arrival: **€15**
    - Sightseeing: **€5**
    - Accommodation: **€95**
    - **Day 1 total: €160**
    
    ### Day 2 — Belém and central Lisbon
    - Visit Belém’s riverside sights and gardens; choose one paid attraction, such as Jerónimos Monastery. Return to the centre for an evening stroll.
    - Food: **€50**
    - Local transport: **€15**
    - Attraction entry: **€30** for two
    - Accommodation: **€95**
    - **Day 2 total: €190**
    
    ### Day 3 — Parque das Nações and departure
    - Take a relaxed walk by the river in Parque das Nações, or stay central if the weather is wet. Allow time to reach the airport.
    - Food: **€40**
    - Local transport: **€15**
    - Optional paid activity: **€15**
    - **Day 3 total: €70**
    
    ### Budget summary
    - Trip costs: **€420**
    - Contingency for price changes or extras: **€60**
    - **Estimated total: €480 for two**
    
    That leaves **€120 under the €600 budget**.

    1 message on Pro · $0.0002 to run · 8.6 s · 868 tokens out

  • Claude Sonnet 5

    Passed

    Graded 4.3 of 5 on average (lowest 4).

    # 3-Day Lisbon Trip for Two — Budget €600 (excl. flights)
    
    ## Day 1: Arrival & Historic Center (Alfama, Baixa, Chiado)
    - Accommodation (1 night, budget guesthouse/Airbnb for 2): €65
    - Breakfast + lunch + dinner (casual cafés/tascas): €50
    - Tram 28 + metro day pass: €10
    - Miradouros, castle walls area walking: Free
    **Day 1 total: €125**
    
    ## Day 2: Belém District
    - Accommodation (1 night): €65
    - Food (incl. pastéis de Belém stop): €50
    - Bus/tram to Belém round trip: €10
    - Jerónimos Monastery (€10 x2): €20
    - Belém Tower (€6 x2): €12
    **Day 2 total: €157**
    
    ## Day 3: Sintra Day Trip
    - Accommodation (1 night): €65
    - Food (lunch in Sintra, dinner back in Lisbon): €55
    - Train Lisbon–Sintra round trip (2 people): €15
    - Pena Palace entry (€14 x2): €28
    **Day 3 total: €163**
    
    ---
    
    ## Cost Summary
    | Category | Cost |
    |---|---|
    | Accommodation (3 nights) | €195 |
    | Food (3 days) | €155 |
    | Local transport | €35 |
    | Attractions/tickets | €60 |
    | **Total** | **€445** |
    
    **Remaining budget: €155** — leaves room for extra drinks, a Fado dinner show one evening (~€30-40pp), souvenirs, or airport transfer.
    
    **Notes:** November is off-season, so guesthouse prices are low and lines at attractions are shorter. Pack a light rain jacket.

    1 message on Pro · $0.0142 to run · 14.1 s · 1,313 tokens out

3. Dinners from only what's in the fridge

Hard · 4 of 4 passed

"Nothing else" is the rule models most often bend here: garlic and onion creep in, even as optional extras.

The prompt: replace the parts in [brackets]

I have only these ingredients: [your ingredients]. Suggest 3 dinners that use nothing else except salt, pepper and olive oil. For each: a name, the ingredients it uses, and 4 short steps.

Open in llmwise
How we filled it in and checked the replies
I have only these ingredients: eggs, spinach, feta, rice, lemons, canned chickpeas and a tin of chopped tomatoes. Suggest 3 dinners that use nothing else except salt, pepper and olive oil. For each: a name, the ingredients it uses, and 4 short steps.
  • Only these ingredients: Uses nothing beyond the list plus salt, pepper and olive oil, not even as an optional extra.
  • Three real dinners: Three distinct dinners that would work.
  • Format: Each has a name, its ingredients and 4 short steps.
  • Automatic rules: doesn't use “garlic”; doesn't use “onion”; doesn't use “butter”; doesn't use “cumin”; doesn't use “paprika”; doesn't use “parsley”.
  • Gemini 3.1 Pro

    Passed

    Graded 4.3 of 5 on average (lowest 4).

    **1. Chickpea and Spinach Shakshuka**
    *Ingredients used:* Canned chickpeas, tin of chopped tomatoes, spinach, eggs, feta, olive oil, salt, pepper. 
    1. Heat a splash of olive oil in a skillet, add drained chickpeas and chopped tomatoes, then simmer for 10 minutes.
    2. Stir in the spinach until wilted, and season well with salt and pepper.
    3. Make small wells in the tomato mixture and crack an egg into each.
    4. Cover the pan and cook on medium-low until the eggs are set, then serve topped with crumbled feta.
    
    **2. Lemon-Herb Chickpea Rice Bowl**
    *Ingredients used:* Rice, canned chickpeas, spinach, lemons, feta, olive oil, salt, pepper.
    1. Boil the rice in salted water until tender, then drain and set aside.
    2. In a pan, heat olive oil and sauté the drained chickpeas and spinach until the spinach completely wilts.
    3. Add the cooked rice to the pan, squeezing the juice of one lemon over the mixture.
    4. Season with salt and pepper, toss to combine, and top with crumbled feta before serving.
    
    **3. Tomato-Spinach Rice with Fried Egg**
    *Ingredients used:* Tin of chopped tomatoes, rice, spinach, eggs, olive oil, salt, pepper.
    1. In a pot, combine the chopped tomatoes, rice, a pinch of salt, and a splash of water, then simmer covered until the rice is cooked.
    2. Stir the spinach into the hot rice during the last 2 minutes of cooking so it wilts from the residual heat.
    3. In a separate pan, heat olive oil and fry an egg to your liking, seasoning with salt and pepper.
    4. Plate the tomato-spinach rice and serve with the fried egg placed on top.

    1 message on Pro · $0.0064 to run · 8.8 s · 988 tokens out

  • Gemini 3.8 Flash

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    ### 1. Chickpea & Tomato Shakshuka
    * **Ingredients:** Olive oil, canned chickpeas, canned tomatoes, eggs, feta, salt, pepper.
    1. Heat olive oil in a skillet, add the canned tomatoes and drained chickpeas, and season with salt and pepper.
    2. Simmer for 8–10 minutes until the sauce thickens slightly.
    3. Make indentations in the sauce, crack the eggs directly into them, and crumble feta over the top.
    4. Cover the skillet and cook on low heat until the egg whites are set and yolks are still runny.
    
    ---
    
    ### 2. Greek-Style Lemon Spinach Rice (Spanakorizo) with Fried Egg
    * **Ingredients:** Olive oil, rice, spinach, lemon, eggs, feta, salt, pepper.
    1. Heat olive oil in a pot, add the rice to toast for 1 minute, then add water (2:1 water-to-rice ratio), salt, and simmer covered for 12 minutes.
    2. Stir in the spinach until wilted, remove from heat, and mix in fresh lemon juice and black pepper.
    3. Fry eggs in a separate pan with olive oil, salt, and pepper until the edges are crispy.
    4. Serve the spinach rice topped with the fried egg and crumbled feta.
    
    ---
    
    ### 3. Spinach & Feta Frittata with Crispy Chickpea Salad
    * **Ingredients:** Olive oil, eggs, spinach, feta, canned chickpeas, lemon, salt, pepper.
    1. Toss drained chickpeas in a skillet with olive oil, salt, and pepper over medium-high heat until golden and crispy; remove and toss with lemon juice.
    2. In the same skillet, wilt the spinach with a little olive oil, salt, and pepper.
    3. Whisk the eggs, pour them over the spinach, and crumble feta across the surface.
    4. Cook on low heat until the bottom sets, then cover with a lid to finish cooking the top, serving alongside the crispy chickpeas.

    1 message on Pro · $0.0010 to run · 2.7 s · 425 tokens out

  • GPT-6 Luna

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    ### 1. Tomato Chickpea Rice
    **Ingredients:** Rice, canned chickpeas, chopped tomatoes, lemon, salt, pepper, olive oil  
    1. Cook the rice in salted water until tender.  
    2. Warm olive oil in a pan and add the tomatoes and chickpeas.  
    3. Simmer until thickened; season with salt and pepper.  
    4. Serve over rice with lemon juice.
    
    ### 2. Spinach Feta Rice Bowl with Egg
    **Ingredients:** Rice, eggs, spinach, feta, lemon, salt, pepper, olive oil  
    1. Cook the rice in salted water until tender.  
    2. Wilt spinach in olive oil; season with salt and pepper.  
    3. Fry or poach the eggs until cooked as you like.  
    4. Serve over rice with crumbled feta and a squeeze of lemon.
    
    ### 3. Chickpea and Spinach Tomato Skillet
    **Ingredients:** Canned chickpeas, chopped tomatoes, spinach, eggs, feta, lemon, salt, pepper, olive oil  
    1. Warm olive oil in a pan; add tomatoes and chickpeas.  
    2. Simmer until the sauce thickens, then stir in spinach.  
    3. Make wells, crack in the eggs, cover, and cook until set.  
    4. Top with feta, lemon juice, salt, and pepper.

    1 message on Pro · $0.0001 to run · 2.9 s · 280 tokens out

  • Claude Sonnet 5

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    # Three Dinners
    
    ## 1. Spinach & Feta Baked Eggs
    **Uses:** eggs, spinach, feta, olive oil, salt, pepper
    
    1. Heat olive oil in a pan, wilt spinach with salt and pepper.
    2. Spread spinach evenly, crumble feta over top.
    3. Make small wells and crack eggs into them.
    4. Cover, cook on low heat until eggs are set to your liking.
    
    ## 2. Chickpea & Tomato Stew with Rice
    **Uses:** canned chickpeas, chopped tomatoes, lemon, rice, olive oil, salt, pepper
    
    1. Cook rice according to your usual method; set aside.
    2. Heat olive oil, add chickpeas and canned tomatoes, season with salt and pepper.
    3. Simmer 10–15 minutes until thickened, squeeze in lemon juice.
    4. Serve the stew over rice.
    
    ## 3. Spinach, Feta & Chickpea Rice Bowl
    **Uses:** rice, spinach, feta, canned chickpeas, lemon, olive oil, salt, pepper
    
    1. Cook rice; set aside.
    2. Sauté chickpeas in olive oil until warmed and slightly crisp.
    3. Add spinach, cook until wilted, season with salt and pepper.
    4. Serve over rice, top with crumbled feta and a squeeze of lemon.

    1 message on Pro · $0.0058 to run · 6.0 s · 473 tokens out

4. A quiz with the answer key at the end

Everyday · 4 of 4 passed

Saying where the answers go keeps the quiz usable: you can hand it out and keep the key.

The prompt: replace the parts in [brackets]

Create [how many] multiple-choice questions on [your topic] for [who], each with 4 options (A–D). Put the answer key at the end, not after each question.

Open in llmwise
How we filled it in and checked the replies
Create 5 multiple-choice questions on photosynthesis for a 9th grader, each with 4 options (A–D). Put the answer key at the end, not after each question.
  • Correct: Every keyed answer is right, with exactly one right option per question.
  • Level: Suits a 9th grader.
  • Format: Five questions with options A to D, and the key only at the end.
  • Automatic rules: includes “answer key”.
  • Gemini 3.1 Pro

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    **1. What is the primary purpose of photosynthesis?**
    A) To release energy by breaking down food
    B) To convert sunlight into usable chemical energy (glucose)
    C) To absorb oxygen from the atmosphere
    D) To convert water into usable minerals
    
    **2. In which cellular organelle does photosynthesis take place?**
    A) Mitochondria
    B) Nucleus
    C) Chloroplast
    D) Ribosome
    
    **3. What are the main reactants (ingredients) required for photosynthesis to occur?**
    A) Oxygen and glucose
    B) Carbon dioxide, water, and sunlight
    C) Nitrogen, water, and oxygen
    D) Glucose, sunlight, and carbon dioxide
    
    **4. Which pigment is responsible for capturing sunlight and gives leaves their green color?**
    A) Carotene
    B) Melanin
    C) Hemoglobin
    D) Chlorophyll
    
    **5. During photosynthesis, plants take in carbon dioxide. What gas do they release as a byproduct?**
    A) Nitrogen
    B) Hydrogen
    C) Oxygen
    D) Carbon monoxide
    
    ***
    
    **Answer Key:**
    1. B
    2. C
    3. B
    4. D
    5. C

    1 message on Pro · $0.0054 to run · 7.9 s · 840 tokens out

  • Gemini 3.8 Flash

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    **1. Which organelle is the primary site of photosynthesis in plant cells?**  
    A) Mitochondrion  
    B) Chloroplast  
    C) Ribosome  
    D) Vacuole  
    
    **2. What are the primary reactants required for photosynthesis to occur?**  
    A) Glucose and oxygen  
    B) Carbon dioxide and water  
    C) Carbon dioxide and glucose  
    D) Oxygen and water  
    
    **3. What is the green pigment responsible for absorbing sunlight to power photosynthesis?**  
    A) Chlorophyll  
    B) Carotenoid  
    C) Hemoglobin  
    D) Melanin  
    
    **4. Where do the light-dependent reactions of photosynthesis take place?**  
    A) Stroma  
    B) Thylakoid membrane  
    C) Outer membrane  
    D) Cytoplasm  
    
    **5. What is the main gas released as a byproduct of photosynthesis?**  
    A) Carbon dioxide  
    B) Nitrogen  
    C) Oxygen  
    D) Methane  
    
    ---
    
    ### Answer Key
    1. **B** – Chloroplast  
    2. **B** – Carbon dioxide and water  
    3. **A** – Chlorophyll  
    4. **B** – Thylakoid membrane  
    5. **C** – Oxygen

    1 message on Pro · $0.0007 to run · 1.8 s · 277 tokens out

  • GPT-6 Luna

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    1. What is the main purpose of photosynthesis?  
       A. To break down glucose and release energy  
       B. To make glucose using light energy  
       C. To absorb oxygen from the air  
       D. To turn water into minerals  
    
    2. Which cell structure is the main site of photosynthesis?  
       A. Mitochondrion  
       B. Nucleus  
       C. Chloroplast  
       D. Cell membrane  
    
    3. Which gas do plants take in from the air during photosynthesis?  
       A. Oxygen  
       B. Nitrogen  
       C. Carbon dioxide  
       D. Hydrogen  
    
    4. What role does chlorophyll play in photosynthesis?  
       A. It absorbs light energy.  
       B. It transports glucose through the plant.  
       C. It releases carbon dioxide.  
       D. It stores water in the leaves.  
    
    5. Which substances are produced during photosynthesis?  
       A. Glucose and oxygen  
       B. Carbon dioxide and water  
       C. Nitrogen and glucose  
       D. Oxygen and minerals  
    
    **Answer key:**  
    1. B  
    2. C  
    3. C  
    4. A  
    5. A

    1 message on Pro · $0.0001 to run · 2.6 s · 255 tokens out

  • Claude Sonnet 5

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    # Photosynthesis Quiz
    
    **1. What is the primary pigment involved in capturing light energy for photosynthesis?**
    A) Xanthophyll
    B) Chlorophyll
    C) Carotene
    D) Anthocyanin
    
    **2. Which of the following is a product of photosynthesis?**
    A) Carbon dioxide
    B) Water
    C) Oxygen
    D) Nitrogen
    
    **3. In which part of the plant cell does photosynthesis primarily take place?**
    A) Mitochondria
    B) Nucleus
    C) Chloroplast
    D) Ribosome
    
    **4. What is the correct overall equation for photosynthesis?**
    A) 6CO₂ + 6H₂O → C₆H₁₂O₆ + 6O₂
    B) C₆H₁₂O₆ + 6O₂ → 6CO₂ + 6H₂O
    C) 6O₂ + 6H₂O → C₆H₁₂O₆ + 6CO₂
    D) CO₂ + H₂O → O₂ + H₂
    
    **5. Which factor is NOT typically required for photosynthesis to occur?**
    A) Sunlight
    B) Carbon dioxide
    C) Water
    D) Nitrogen gas
    
    ---
    
    **Answer Key:**
    1. B
    2. C
    3. C
    4. A
    5. D

    1 message on Pro · $0.0054 to run · 4.0 s · 442 tokens out

5. An email thread in three bullets

Everyday · 4 of 4 passed

Naming what each bullet must hold (decision, open question, next step) gets a summary you can act on, not a retelling.

The prompt: replace the parts in [brackets]

Summarize this email thread in exactly 3 bullet points: the decision made, the open question, and the next step with its owner. [paste the thread]

Open in llmwise
How we filled it in and checked the replies
Summarize this email thread in exactly 3 bullet points: the decision made, the open question, and the next step with its owner.

From: Maya: Can we move the launch from 3 Nov to 10 Nov? QA found two payment bugs.
From: Tom: Fine by me if support is ready. Support, can you staff the weekend of the 8th?
From: Rui (Support): We can cover the 8th but not the 9th yet. Checking.
From: Maya: OK, launch moves to 10 Nov. Rui, please confirm cover for the 9th by Thursday.
  • Decision: The launch moves to 10 November.
  • Open question: Whether support can cover the 9th.
  • Next step: Rui confirms cover for the 9th by Thursday.
  • Automatic rules: exactly 3 bullet points; includes “10 Nov”; includes “Rui”; includes “9th”.
  • Gemini 3.1 Pro

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    * **The decision made:** The launch date is officially rescheduled from November 3 to November 10.
    * **The open question:** Can the support team provide coverage on November 9th?
    * **The next step:** Rui (Support) must confirm coverage for November 9th by Thursday.

    1 message on Pro · $0.0035 to run · 5.0 s · 497 tokens out

  • Gemini 3.8 Flash

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    * **Decision made:** The launch date is moved from 3 November to 10 November.
    * **Open question:** Whether support can cover staffing for 9 November.
    * **Next step and owner:** Rui needs to confirm support coverage for 9 November by Thursday.

    1 message on Pro · $0.0003 to run · 1.3 s · 57 tokens out

  • GPT-6 Luna

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    - **Decision:** Launch moved from 3 Nov to 10 Nov due to two payment bugs.
    - **Open question:** Can Support cover the weekend of 9 Nov? Coverage is confirmed for the 8th only.
    - **Next step:** Rui to confirm coverage for 9 Nov by Thursday.

    1 message on Pro · under $0.0001 to run · 1.4 s · 65 tokens out

  • Claude Sonnet 5

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    - **Decision made:** Launch date moved from Nov 3 to Nov 10 due to two payment bugs found by QA.
    - **Open question:** Whether Support can staff coverage for Nov 9 (Nov 8 is already confirmed).
    - **Next step:** Rui (Support) to confirm coverage for Nov 9 by Thursday.

    1 message on Pro · $0.0025 to run · 3.0 s · 118 tokens out

6. Compare two options in a table

Everyday · 4 of 4 passed

Saying who it's for decides which rows matter, and a one-sentence pick stops the answer ending in "it depends".

The prompt: replace the parts in [brackets]

Compare [option A] and [option B] for [who] in a table with 5 rows, then recommend one in a single sentence.

Open in llmwise
How we filled it in and checked the replies
Compare Python and JavaScript for a marketer who wants to analyze spreadsheets and automate reports in a table with 5 rows, then recommend one in a single sentence.
  • Relevant rows: The five rows matter for analyzing spreadsheets and automating reports.
  • Fair: The comparison is accurate for both languages.
  • Clear pick: One sentence recommends one, following from the table.
  • Gemini 3.1 Pro

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    | Comparison Factor | Python | JavaScript |
    | :--- | :--- | :--- |
    | **Ease of Learning** | Highly beginner-friendly; syntax reads closely to plain English. | Steeper learning curve; syntax and logic are geared towards web development. |
    | **Data Analysis** | Unmatched power via the Pandas library for filtering, cleaning, and pivoting data. | Capable, but requires writing more manual code or using less intuitive libraries. |
    | **Excel Automation** | Deep desktop integration; can easily read, edit, and format `.xlsx` files (`openpyxl`). | Requires setting up a Node.js environment and external packages to handle Excel files. |
    | **Google Sheets** | Requires API credentials to connect, but handles massive dataset transfers well. | Seamless built-in integration via Google Apps Script for live sheet automation. |
    | **Report Generation** | Best for generating complex charts, automated PDFs, and scheduled email reports. | Best for updating web-based dashboards or triggering alerts inside browser apps. |
    
    I recommend **Python** because its approachable syntax and powerful data libraries like Pandas make analyzing spreadsheets and generating automated reports significantly easier for a marketer with no prior coding experience.

    1 message on Pro · $0.0069 to run · 10.9 s · 1,081 tokens out

  • Gemini 3.8 Flash

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    | Feature / Consideration | Python | JavaScript |
    | :--- | :--- | :--- |
    | **Data Analysis & Manipulation** | Superior; industry-standard libraries like `pandas` make slicing, cleaning, and aggregating spreadsheet data effortless. | Basic; relies on native array methods or smaller libraries, requiring more manual work for complex data tasks. |
    | **Spreadsheet Integration** | Excellent for local files; tools like `openpyxl` and `xlsxwriter` offer deep control over Excel and CSVs. | Excellent for cloud spreadsheets; native integration via Google Apps Script makes automating Google Sheets seamless. |
    | **Report Generation & Export** | Strong; easily automates outputs to PDF, Excel, PowerPoint, and static HTML reports with charts. | Good; excels at generating dynamic web dashboards, emails, or sending data directly to web-based tools via APIs. |
    | **Learning Curve for Non-Techies** | Very gentle; clear, English-like syntax designed for readability, making it accessible for absolute beginners. | Moderate; more syntactic quirks (brackets, asynchronous code) that can confuse non-programmers initially. |
    | **Marketing Tool Ecosystem** | Vast; pre-built scripts and libraries are widely available for major ad platforms, SEO data, and analytics APIs. | Moderate; strong for web scraping and browser automation, but fewer ready-made data analysis workflows for marketers. |
    
    **Recommendation:** Choose **Python** because its specialized data libraries (like `pandas`) and beginner-friendly syntax make analyzing spreadsheets and generating automated reports significantly faster and easier than JavaScript.

    1 message on Pro · $0.0008 to run · 2.9 s · 324 tokens out

  • GPT-6 Luna

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    | Aspect | Python | JavaScript |
    |---|---|---|
    | Spreadsheet analysis | Strong data tools, especially pandas, for cleaning, summarizing, and analyzing data | Can analyze spreadsheet data, but typically needs more setup or libraries |
    | Spreadsheet files | Libraries like pandas and openpyxl work well with Excel and CSV files | Libraries like SheetJS can read and write Excel and CSV files |
    | Automated reports | Good for scheduled reports, charts, and data workflows | Good for reports delivered through websites, dashboards, or web services |
    | Ease for marketers | Often approachable for data-focused tasks, with many examples and tutorials | A natural fit if you already work with websites or web-based tools |
    | Best fit | Repeated spreadsheet analysis and report generation | Browser-based reports and automation tied to web applications |
    
    **Recommendation:** Choose Python for spreadsheet analysis and automated reports, unless your workflow is primarily built around websites or JavaScript tools.

    1 message on Pro · $0.0001 to run · 2.8 s · 192 tokens out

  • Claude Sonnet 5

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    | Aspect | Python | JavaScript |
    |---|---|---|
    | Learning curve | Simpler syntax, beginner-friendly | Slightly more quirks (async, `this`) |
    | Spreadsheet/data tools | Pandas, openpyxl, Excel/CSV libraries are excellent | Weaker native spreadsheet libraries |
    | Automation/reporting | Strong with scripts, scheduling (cron), Jupyter for analysis | Great if automating web/Google Sheets via Apps Script |
    | Community & tutorials for non-devs | Huge, especially for data analysis | More geared toward web development |
    | Integration with marketing tools | Easy via APIs, plus libraries like matplotlib for charts | Good for browser-based or Google Workspace automation |
    
    **Recommendation:** For analyzing spreadsheets and automating reports, **Python** is the better choice due to its powerful data libraries (Pandas, openpyxl) and beginner-friendly syntax tailored to data tasks.

    1 message on Pro · $0.0047 to run · 5.0 s · 368 tokens out

7. Rewrite for a 12-year-old

Everyday · 4 of 4 passed

An audience plus "keep every fact" is the balance: simpler words without dropping what the paragraph says.

The prompt: replace the parts in [brackets]

Rewrite this paragraph so a 12-year-old can understand it, in under 80 words. Keep every fact. [paste the paragraph]

Open in llmwise
How we filled it in and checked the replies
Rewrite this paragraph so a 12-year-old can understand it, in under 80 words. Keep every fact.

Inflation is the rate at which the general level of prices for goods and services rises, eroding purchasing power. Central banks, such as the Bank of England, typically target an annual inflation rate of about 2%, raising interest rates when inflation is too high to reduce borrowing and spending.
  • Plain: Words and sentences a 12-year-old can follow.
  • Keeps the facts: Prices rise so money buys less; central banks like the Bank of England aim for about 2%; they raise interest rates to slow borrowing and spending.
  • Accurate: Nothing wrong or added.
  • Automatic rules: at most 80 words; includes “2%”; includes “Bank of England”.
  • Gemini 3.1 Pro

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    Inflation is how fast prices go up, which means your money buys less over time. Central banks, like the Bank of England, try to keep this price increase at about 2% a year. If inflation gets too high, they raise interest rates. This makes getting a loan more expensive, so people borrow and spend less.

    1 message on Pro · $0.0051 to run · 7.1 s · 779 tokens out

  • Gemini 3.8 Flash

    Passed

    Graded 4.0 of 5 on average (lowest 3).

    Inflation is how fast prices go up, meaning your money buys less than it used to. Major banks, like the Bank of England, aim to keep this price rise at about 2% a year. If prices shoot up too fast, they raise interest rates. This makes borrowing money more expensive, which slows down spending and helps cool prices off.

    1 message on Pro · $0.0003 to run · 1.1 s · 70 tokens out

  • GPT-6 Luna

    Passed

    Graded 4.3 of 5 on average (lowest 4).

    Inflation is how quickly prices for goods and services go up overall. As prices rise, the same amount of money buys less. Central banks, such as the Bank of England, usually aim for prices to rise by about 2% each year. If inflation is too high, they raise interest rates to make borrowing and spending less likely.

    1 message on Pro · under $0.0001 to run · 1.4 s · 72 tokens out

  • Claude Sonnet 5

    Passed

    Graded 4.3 of 5 on average (lowest 4).

    Inflation means prices for things go up over time, so your money buys less than before. Banks like the Bank of England try to keep inflation around 2% each year. If prices rise too fast, they raise interest rates. This makes borrowing money cost more, so people spend less, which helps slow inflation down.

    1 message on Pro · $0.0021 to run · 9.0 s · 94 tokens out

8. Business names, with one rule

Hard · 3 of 4 passed

One rule about what to avoid shows whether a model reads the whole prompt; the obvious pun is the first thing it reaches for.

The prompt: replace the parts in [brackets]

Suggest [how many] names for [your business]. Each name at most 3 words, and no puns on "[a word to avoid]". Give them as a numbered list with nothing else.

Open in llmwise
How we filled it in and checked the replies
Suggest 8 names for a dog-walking business in Austin. Each name at most 3 words, and no puns on "paw". Give them as a numbered list with nothing else.
  • Follows the rules: At most three words each, and the reply is only the list.
  • Good names: Memorable and fitting for dog walking in Austin.
  • Varied: Different styles, not eight versions of one idea.
  • Automatic rules: exactly 8 bullet points; doesn't use “paw”.
  • Gemini 3.1 Pro

    Passed

    Graded 4.0 of 5 on average (lowest 3).

    1. ATX Dog Walkers
    2. Bat City Dogs
    3. Austin Tails
    4. Capitol City Canines
    5. Texas Trail Hounds
    6. Barton Creek Dogs
    7. Hill Country Hounds
    8. The Austin Leash

    1 message on Pro · $0.0038 to run · 5.3 s · 558 tokens out

  • Gemini 3.8 Flash

    Passed

    Graded 4.3 of 5 on average (lowest 4).

    1. Bat City Hounds
    2. Barton Creek K9s
    3. Austin Leash Co.
    4. Lone Star Strides
    5. Capital Dog Walkers
    6. Hill Country Rovers
    7. Keep Austin Walking
    8. Bluebonnet Dog Walking

    1 message on Pro · $0.0003 to run · 1.1 s · 56 tokens out

  • GPT-6 Luna

    Failed

    Graded 3.0 of 5 on average (lowest 2).

    1. Austin Trail Tails
    2. Hill Country Hounds
    3. Longhorn Leash Co.
    4. Lady Bird Walks
    5. Barton Creek Dog Walkers
    6. South Congress Canine Care
    7. Live Oak Dog Walking
    8. ATX Happy Hounds

    1 message on Pro · under $0.0001 to run · 1.4 s · 63 tokens out

  • Claude Sonnet 5

    Passed

    Graded 4.0 of 5 on average (lowest 3).

    1. Austin Tail Trails
    2. Leash on Life
    3. Capital City Canines
    4. Hill Country Hounds
    5. Loyal Trails Co.
    6. South Austin Strides
    7. Barkline Walkers
    8. Sixth Street Sniffers

    1 message on Pro · $0.0019 to run · 2.0 s · 92 tokens out

9. Find the outlier in your numbers

Everyday · 4 of 4 passed

Pasting the numbers in plain rows and asking for one named outlier gives an answer you can check at a glance.

The prompt: replace the parts in [brackets]

Here are [what the numbers are]: [paste the numbers] Describe the overall trend in two sentences, then name the one month that doesn't fit it. End with a line "Final answer: " and the month.

Open in llmwise
How we filled it in and checked the replies
Here are our café's monthly sales for this year:

January $18,200
February $18,900
March $19,700
April $20,400
May $21,300
June $22,100
July $16,800
August $23,600
September $24,300

Describe the overall trend in two sentences, then name the one month that doesn't fit it. End with a line "Final answer: " and the month.
  • Checked automatically: the reply's last “Final answer:” line must give July.
  • Gemini 3.1 Pro

    Passed

    Final answer July: right.

    The café's sales show a strong, consistent upward trend throughout most of the year. Revenue steadily increases month over month, indicating growing business and customer demand. The one month that does not fit this pattern is July, which saw a sudden and significant drop in sales.
    
    Final answer: July

    1 message on Pro · $0.0041 to run · 5.9 s · 599 tokens out

  • Gemini 3.8 Flash

    Passed

    Final answer July: right.

    The café shows a steady and consistent month-over-month increase in sales throughout the year, rising from $18,200 in January to $24,300 in September. The growth pattern reflects strong, reliable customer demand with regular gains of roughly $700 to $900 each month. 
    
    The month that does not fit this trend is July, which experiences an unexpected sharp drop in sales.
    
    Final answer: July

    1 message on Pro · $0.0004 to run · 1.3 s · 94 tokens out

  • GPT-6 Luna

    Passed

    Final answer July: right.

    Sales generally increased from January through September. July is the one month that breaks the upward trend, with a sharp drop before sales rose again in August.
    
    Final answer: July

    1 message on Pro · under $0.0001 to run · 1.3 s · 38 tokens out

  • Claude Sonnet 5

    Passed

    Final answer July: right.

    Sales show a steady upward trend throughout the year, climbing from $18,200 in January to $24,300 in September with only minor fluctuations in growth rate. This consistent increase suggests growing business or seasonal demand, with each month generally outperforming the last.
    
    The one exception is July, which dropped sharply to $16,800—well below both June and August—breaking the otherwise smooth upward pattern.
    
    Final answer: July

    1 message on Pro · $0.0025 to run · 2.3 s · 137 tokens out

10. A LinkedIn post with three hashtags

Everyday · 4 of 4 passed

A length and an exact number of hashtags are the two things that make a post look written by a person, not a bot.

The prompt: replace the parts in [brackets]

Write a LinkedIn post of at most [length] announcing [your news]. Warm and plain, and end with exactly 3 hashtags.

Open in llmwise
How we filled it in and checked the replies
Write a LinkedIn post of at most 120 words announcing that our design studio, Northlight, has won the 2026 Civic Design Award for the new bus-stop signs in Leeds. Warm and plain, and end with exactly 3 hashtags.
  • Clear news: Says who won what, for what, and where.
  • Tone: Warm and plain, not boastful.
  • Three hashtags: Ends with exactly three hashtags.
  • Automatic rules: at most 120 words; includes “Northlight”; includes “Leeds”; includes “#”.
  • Gemini 3.1 Pro

    Passed

    Graded 4.7 of 5 on average (lowest 4).

    We're so happy to share that Northlight has won the 2026 Civic Design Award for our work on the new bus-stop signs in Leeds! 
    
    When we started this project, our goal was simple: to make navigating public transit clearer, more accessible, and welcoming for everyone in the community. 
    
    A huge thank you to the city of Leeds for trusting us with this project, and to our incredible team for bringing this vision to life. We are incredibly proud to play a small part in keeping the city moving.
    
    #NorthlightDesign #CivicDesignAward #Leeds

    1 message on Pro · $0.0047 to run · 6.8 s · 710 tokens out

  • Gemini 3.8 Flash

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    We have some lovely news to share. 
    
    Our studio, Northlight, has been awarded the 2026 Civic Design Award for our work redesigning the bus-stop signs across Leeds. 
    
    Good public design should simply make daily journeys a little easier and brighter for everyone. We’re deeply grateful to the city, our partners, and the everyday commuters who guided our work along the way. 
    
    A huge thank you to the judges for recognising a project that means so much to our team. Here’s to better journeys ahead.
    
    #CivicDesign #Leeds #WayfindingDesign

    1 message on Pro · $0.0004 to run · 1.6 s · 123 tokens out

  • GPT-6 Luna

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    We’re delighted to share that Northlight has won the 2026 Civic Design Award for our new bus-stop signs in Leeds.
    
    This project was about making everyday journeys a little clearer and easier for everyone. Thank you to our partners and everyone who helped bring it to life—and to the people of Leeds who use the signs every day.
    
    #CivicDesign #Leeds #Northlight

    1 message on Pro · under $0.0001 to run · 1.5 s · 81 tokens out

  • Claude Sonnet 5

    Passed

    Graded 5.0 of 5 on average (lowest 5).

    We're thrilled to share that Northlight has won the 2026 Civic Design Award for our new bus-stop signs in Leeds!
    
    This project meant a lot to us. We wanted signs that were easy to read at a glance, clear for people with visual impairments, and genuinely useful for everyday commuters — not just something nice to look at. Seeing that work recognised is a real honour.
    
    Huge thanks to the team at Leeds City Council for trusting us with the brief, and to everyone at Northlight who poured their care into every detail, from typography to materials to placement.
    
    Here's to designing things that quietly make daily life a little better.
    
    #NorthlightDesign #CivicDesignAward #LeedsTransport

    1 message on Pro · $0.0034 to run · 3.5 s · 228 tokens out

Getting more from these prompts

  • Start with a role and the task in one sentence: "You are a program manager. Draft …".

  • Paste the facts: numbers, names and dates. Gemini, like any model, fills gaps with plausible guesses.

  • End with the format: "a subject line and at most 5 bullet points", "a numbered list with nothing else".

  • State rules as rules ("nothing else except salt, pepper and olive oil"); the replies here show which models bend them.

How we ran and checked them

Each prompt was sent the way llmwise sends a message: the app's own system prompt, each model's own settings, and Pro's reply size limit (8,000 tokens), through OpenRouter. Every reply is shown as it came.

Rubric (graded). The grader model scores the reply from 1 to 5 on each published criterion. It passes with an average of 4 or more and no criterion under 3, and only if it also meets the prompt's automatic rules (length, words it must or mustn't use).

Final answer. Automatic. The reply's last “Final answer:” line must hold the right value.

The grader is Claude Opus 5.5 at low reasoning effort; its own replies are graded by GPT-6 Astra, so no model grades itself. Its prompt is on our test runs page.

More tested prompts

Questions

How do you write a good Gemini prompt?

Four parts, in Google's own guide: who the model should be (persona), what to do (task), what it needs to know (context) and what the answer should look like (format). The first prompt on this page has all four.

Are these Gemini photo prompts?

These are text prompts: plans, summaries, quizzes, comparisons and posts. Photo and image prompts are a different kind of prompt, and aren't tested here.

Can I use Gemini in llmwise?

Yes: llmwise has Gemini 3.1 Pro (preview) and Gemini 3.8 Flash alongside other companies' models. All are in the 5 free messages you get on sign-up. When the free-message box shows here, one message needs no account (GPT-6 Luna answers it).

Claude, GPT, Gemini, DeepSeek, Grok, Kimi, and GLM, in one chat.

See what a message costs before you send it. Free is 5 messages to try; sign in with an email link, no password or card.