Skip to content

Tested prompt · AI poem generator

A sonnet: every AI model's reply, tested

We ran this hard AI poem generator test input on all 19 models in llmwise, the way the tool runs it, and checked every reply the same way. Here's each one as it came, with whether it passed, what it cost and how long it took.

Based on 19 of our test runs on , through OpenRouter with the tool's own instructions and the app's settings. Updated .

Short answer

13 of 19 models passed this AI poem generator test input's check (rubric); Claude Sonnet 5.5, Claude Sonnet 5, Claude Haiku 4.5, Gemini 3.1 Pro, GLM 5.3, and Mistral Large 4 didn't. The cheapest reply that passed was GLM 5.3 Flash's, at $0.00014, and the fastest too, in 5.5 s. The dearest reply, Claude Fable 5.1's, cost 594 times as much ($0.0855).

The prompt, as sent, and its check

Checked by rubric (graded), the same way for every model.

A sonnet (hard)

Write an original poem about this, with a title on the first line. Keep every name and detail in it. Write it as a sonnet.

---
Learning to cook my mother's recipes after she moved away.

Sent with the AI poem generator's own instructions as the system prompt, as a free run of the tool sends them.

Graded on:

  • Form: Apart from the title, it has fourteen lines of roughly ten syllables in iambic metre, and ends in a rhyming couplet.
  • Concrete: It's built from specific images of cooking (a dish, a smell, a tool), not general statements about missing someone.
  • A turn: There's a turn in thought or feeling, as a sonnet has, and the couplet closes it.

How the AI poem generator's test inputs are scored, with every model's results on all five.

Every model's result

All 19 models on this test input, in catalog order.

Every model's reply to “A sonnet”
ModelResultCostTimeReply
Claude Fable 5.1AnthropicPassed: Graded 5.0 of 5 on average (lowest 5).$0.085524.1 s235 tokens
Claude Opus 5.5AnthropicPassed: Graded 5.0 of 5 on average (lowest 5).$0.032817.7 s244 tokens
Claude Sonnet 5.5AnthropicFailed: Graded 3.7 of 5 on average (lowest 3).$0.00435.1 s327 tokens
Claude Sonnet 5AnthropicFailed: Graded 3.7 of 5 on average (lowest 3).$0.00405.0 s230 tokens
Claude Haiku 5.5AnthropicPassed: Graded 4.7 of 5 on average (lowest 4).$0.001619.3 s249 tokens
Claude Haiku 4.5AnthropicFailed: Graded 3.3 of 5 on average (lowest 3).$0.00123.0 s166 tokens
GPT-6 AstraOpenAIPassed: Graded 4.7 of 5 on average (lowest 4).$0.0316122.7 s157 tokens
GPT-6.1 SolOpenAIPassed: Graded 4.3 of 5 on average (lowest 4).$0.004915.2 s160 tokens
GPT-6 SolOpenAIPassed: Graded 4.0 of 5 on average (lowest 3).$0.014932.0 s155 tokens
GPT-6 LunaOpenAIPassed: Graded 4.0 of 5 on average (lowest 3).$0.0003613.6 s147 tokens
Gemini 3.1 Pro (preview)GoogleFailed: Graded 3.7 of 5 on average (lowest 2).$0.021928.5 s144 tokens
Gemini 3.8 FlashGooglePassed: Graded 4.0 of 5 on average (lowest 3).$0.00267.6 s141 tokens
DeepSeek V4.1 FlashDeepSeekPassed: Graded 4.0 of 5 on average (lowest 4).$0.007818.2 s153 tokens
DeepSeek V4 ProDeepSeekPassed: Graded 4.0 of 5 on average (lowest 3).$0.018734.4 s158 tokens
Grok 4.7xAIPassed: Graded 4.0 of 5 on average (lowest 3).$0.057087.9 s143 tokens
Kimi K3MoonshotPassed: Graded 4.0 of 5 on average (lowest 4).$0.00227.6 s162 tokens
GLM 5.3Z.aiFailed: Graded 3.3 of 5 on average (lowest 2).$0.00111.7 s156 tokens
GLM 5.3 FlashZ.aiPassed: Graded 4.0 of 5 on average (lowest 3).$0.000145.5 s161 tokens
Mistral Large 4MistralFailed: No answer within Pro's reply limit of 8,000 tokens: the model spent them all reasoning.$0.0170109.5 s1,851 tokens

Cost: what OpenRouter charged us for the reply. Time: from sending to the whole reply. Reply: its length in tokens, thinking not counted. In llmwise you pay per message, not per token: each of these replies counted as one message on Pro.

Every reply

The 6 replies that failed first, then the ones that passed, from the cheapest up.

  1. GLM 5.3

    Failed: Graded 3.3 of 5 on average (lowest 2).

    A Sonnet for Her Kitchen
    
    Her recipe cards are smudged with old tomato,
    written in her hand the year I learned to walk.
    Now she has moved three hundred miles away,
    and I am left to decipher what she wrote—
    a pinch of this, a swirl of that, of cumin,
    the bread that rises while the rain comes down.
    I burn the garlic twice before I learn
    her trick of lowering flame and breathing slow.
    Each Sunday I call her with a floury hand
    and ask about the soup she made in March.
    She laughs and says the secret's only time,
    the same slow simmer she would let me stir
    when I was small and standing on a stool.
    The pot hums on. Her kitchen blooms in mine.

    377 tokens in, 156 out · 1.7 s · $0.0011 · 1 message on Pro · answered by z-ai/glm-5.3 via Friendli ·

  2. Claude Haiku 4.5

    Failed: Graded 3.3 of 5 on average (lowest 3).

    Learning My Mother's Recipes
    
    I stand before the stove she used to know,
    with flour dusting counters, pots and pans,
    the handwritten card in her careful hand's slow
    account of measurements and shifting plans.
    
    The onions sizzle in the heavy skillet
    the way she showed me once, before she went,
    her voice still here though distance now can fill it,
    each step a small and sacred sacrament.
    
    I taste and taste again, adjusting salt,
    trying to chase the flavor of her kitchen,
    the way her laughter rose without a halt
    above the steam that made the windows glisten.
    
    These recipes are love that will not fade—
    the meals we'll share, the distance that she made.

    386 tokens in, 166 out · 3.0 s · $0.0012 · 1 message on Pro · answered by anthropic/claude-haiku-4.5 via Anthropic ·

  3. Claude Sonnet 5

    Failed: Graded 3.7 of 5 on average (lowest 3).

    The Recipes She Left
    
    I stand where she once stood beside the stove,
    and try to guess the measure of her hand—
    a pinch, a dash, the things she never wrote,
    now that my mother's moved to farther land.
    
    I call her up when garlic starts to burn,
    she laughs and says, "you're rushing it again."
    Each dish I make becomes a slow return
    to kitchens I once knew before, back then.
    
    The broth won't taste the way that hers once did,
    I miss a spice I cannot quite recall—
    but flavor by mistaken flavor, hid
    inside my clumsy hands, her voice stands tall.
    
    So through these pots I learn to hold her near:
    her recipes rebuild what moved from here.

    496 tokens in, 304 out (74 of them reasoning) · 5.0 s · $0.0040 · 1 message on Pro · answered by anthropic/claude-sonnet-5 via Anthropic ·

  4. Claude Sonnet 5.5

    Failed: Graded 3.7 of 5 on average (lowest 3).

    Her Recipes, My Kitchen
    
    I keep her stained index card by the stove,
    the ink gone soft where thumbs of hers have pressed.
    It says "some salt," "enough," "until it's Rove"—
    
    I need to fix that line; here is the poem:
    
    Her Recipes, My Kitchen
    
    I prop her stained recipe card by the stove,
    its ink gone soft where all her thumbs have pressed.
    It says "add salt," then "just enough," then "rove
    the spoon around until it looks right." Guessed
    at, every measure lives inside her hands.
    I brown the onions slower than I should,
    and watch the pan for what she understands:
    the hush before the sweetness, sharp turned good.
    The kitchen smells almost like hers, but thin,
    a song remembered with the middle gone.
    I call her up. She laughs and says, "Begin
    with patience, love." I stir the whole night on.
       The distance holds us, simmering, apart,
       yet every spoonful tastes of her, and heart.

    498 tokens in, 327 out · 5.1 s · $0.0043 · 1 message on Pro · answered by anthropic/claude-sonnet-5.5 via Anthropic ·

  5. Mistral Large 4

    Failed: No answer within Pro's reply limit of 8,000 tokens: the model spent them all reasoning.

    (no reply)

    372 tokens in, 8,000 out (6,149 of them reasoning) · 109.5 s · $0.0170 · 1 message on Pro · answered by mistralai/mistral-large-4-0 via Mistral ·

  6. Gemini 3.1 Pro

    Failed: Graded 3.7 of 5 on average (lowest 2).

    Simmering Memories
    
    The kitchen feels so empty and so bare,
    Since mother moved away to change her life.
    I gather up her recipes with care,
    And grip the handle of her chopping knife.
    
    I learn to cook the dishes that she made,
    With stains of olive oil and dark red wine.
    I watch the garlic in the skillet fade,
    And hope to make the complex flavors mine.
    
    I read her looping script in faded pen,
    And measure yeast and water for the bread.
    I try to recreate the meals again,
    Remembering the helpful things she said.
    
    Although she lives a thousand miles apart,
    I cook her food to keep her in my heart.

    376 tokens in, 3,595 out (3,451 of them reasoning) · 28.5 s · $0.0219 · 1 message on Pro · answered by google/gemini-3.1-pro-preview via Google AI Studio ·

  7. GLM 5.3 Flash

    Passed: Graded 4.0 of 5 on average (lowest 3).

    The Kitchen After You've Gone
    
    The recipe card is written in your hand,
    each note a small instructions map of home:
    a pinch of this, remember, stir, don't rush
    the roux until it turns the shade of dark amber.
    
    I measure flour you never had to weigh,
    your wrists remembering what mine must learn.
    The phone rings late; you walk me through the stock,
    your voice across the distance like the warmth
    that used to fill the kitchen when we cooked
    side by side, you chopping, me on stirring duty.
    
    Now steam fogs windows in a stranger's town,
    my pot holds something closer than the soup:
    each time I get it right, I set a place
    for you, and feed the distance with your taste.

    377 tokens in, 175 out (14 of them reasoning) · 5.5 s · $0.0001 · 1 message on Pro · answered by z-ai/glm-5.3-flash via Friendli ·

  8. GPT-6 Luna

    Passed: Graded 4.0 of 5 on average (lowest 3).

    Her Recipes
    
    When Mother moved away, I stayed to learn
    Her recipes in ink beside the flame
    The onions softly hiss and start to turn
    I miss the hand that taught each spice its name
    
    I measure salt as if it were her voice
    The wooden spoon taps time against the pan
    I taste, then add, and make the careful choice
    To trust the notes and do the best I can
    
    The simmer lifts its silver veil aloft
    The sauce grows dark beneath my watchful eye
    The flavors settle, warm and deep and soft
    I send one wish along the evening sky
    
    Though she has moved away, her flavors stay
    I learn her recipes, and find my way.

    367 tokens in, 1,360 out (1,213 of them reasoning) · 13.6 s · $0.0004 · 1 message on Pro · answered by openai/gpt-6-luna via OpenAI ·

  9. Claude Haiku 5.5

    Passed: Graded 4.7 of 5 on average (lowest 4).

    Her Handwriting, My Stove
    
    The card is spotted, softened at the fold,
    her looping ink says "salt until it's right,"
    and "onions slow," as she has always told.
    I stand where she once stood, the kitchen light
    
    thin on the sill. I cook alone, lifting each pot,
    and let the onions hiss until they brown,
    then taste, and add, and taste, and wonder what
    she left out, the part she never wrote down.
    
    Her town is four hours off, the cord, the phone.
    She calls to hear the sizzle, hums a rhyme,
    and laughs when garlic burns, "You're not alone."
    Scrape off the black; the rest will do in time.
    
    I plate two portions, set hers across from mine,
    and eat the sauce she taught me, line by line.

    498 tokens in, 3,015 out (2,766 of them reasoning) · 19.3 s · $0.0016 · 1 message on Pro · answered by anthropic/claude-haiku-5.5 via Anthropic ·

  10. Kimi K3

    Passed: Graded 4.0 of 5 on average (lowest 4).

    Her Recipes, My Hands
    
    She moved away and left the kitchen bare,
    save for the cards her careful hand had penned,
    each recipe a map to lead me there,
    back to the table where our talks would end.
    I measured flour the way she used to do,
    a pinch, a handful, never quite precise,
    and burned the onions first, then tried anew,
    and learned the simmer from her small advice.
    The phone would ring mid-stir and she would say,
    "taste it, you'll know," and somehow I would know,
    her voice a spoon that stirred from miles away,
    her patient love still teaching soft and slow.
    Now when the steam rises warm above the pan,
    she's here again, though far, as no one else quite can.

    455 tokens in, 222 out (60 of them reasoning) · 7.6 s · $0.0022 · 1 message on Pro · answered by moonshotai/kimi-k3 via Wafer ·

  11. Gemini 3.8 Flash

    Passed: Graded 4.0 of 5 on average (lowest 3).

    Inherited Flavors
    
    The kitchen holds the quiet that you left,
    Your handwritten cards upon the counter lie;
    Of your warm presence suddenly bereft,
    I watch the butter melt and softly sigh.
    
    The yellow paper stained with grease and tea,
    Each faded note instructs me where to start;
    Across the miles that keep you far from me,
    I learn the measured secrets of your art.
    
    The bubbling broth begins to fill the room
    With rosemary and garlic rich and deep,
    Dispersing all the distance and the gloom,
    Awaking memories I longed to keep.
    
    Though miles divide us since you moved away,
    Your warmth and flavor guide my hands today.

    376 tokens in, 1,335 out (1,194 of them reasoning) · 7.6 s · $0.0026 · 1 message on Pro · answered by google/gemini-3.8-flash via Google AI Studio ·

  12. GPT-6.1 Sol

    Passed: Graded 4.3 of 5 on average (lowest 4).

    Since Mother Moved Away
    
    I learn to cook the meals my mother made;
    I prop her book beside the hissing flame.
    The kitchen wears her absence like a shade;
    I taste the soup and whisper out her name.
    
    For now, I burn the bread and bruise the thyme;
    The onions catch; I scrape the pan too late.
    Her penciled notes say, Give the dough more time.
    I let it rise, and teach my hands to wait.
    
    She moved away; I cannot bridge the space,
    But butter melts the way it did before.
    I lean above the steam to find her face;
    Her voice comes back: A pinch, and nothing more.
    
    The meal is mine, though hers in every part;
    Each dish becomes a lesson for my heart.

    367 tokens in, 899 out (739 of them reasoning) · 15.2 s · $0.0049 · 1 message on Pro · answered by openai/gpt-6.1-sol via OpenAI ·

  13. DeepSeek V4.1 Flash

    Passed: Graded 4.0 of 5 on average (lowest 4).

    Learning My Mother's Recipes After She Moved Away
    
    The old kitchen holds her handwriting still,
    a splattered card of butter, salt, and flour.
    I measure cinnamon against my will,
    and stir the pot through half a lonely hour.
    
    She moved away; the kitchen still is hers.
    The cinnamon, the bay leaf, and the thyme
    sit patient on the shelf, and something stirs
    I see her hands at every supper time.
    
    I call her up; she answers on the phone.
    "How much is salt?" I ask. She laughs, her voice
    a warm instruction I have always known.
    I stir the pot and make a silent choice.
    
    to keep her near in a simmering spoon,
    and find her home inside a quiet noon.

    386 tokens in, 6,497 out (6,344 of them reasoning) · 18.2 s · $0.0078 · 1 message on Pro · answered by deepseek/deepseek-v4.1-flash via Together ·

  14. GPT-6 Sol

    Passed: Graded 4.0 of 5 on average (lowest 3).

    The First Loaf
    I read the recipes she left behind.
    My mother moved away; I learn to cook.
    I hear her voice, though only in my mind;
    Her penciled margins tell me where to look.
    
    The onions hiss; I lower down the heat.
    The garlic browns too fast inside the pan.
    I scrape it out, concede a small defeat,
    Then start again and do the best I can.
    
    Her old blue bowl sits waiting in the light.
    I knead the dough and let it slowly rise.
    I set one plate; the kitchen dims to night.
    I watch the loaf through glass with hopeful eyes.
    
    She lives too far to taste the bread I made,
    But in its warmth, her care has come and stayed.

    367 tokens in, 2,908 out (2,753 of them reasoning) · 32.0 s · $0.0149 · 1 message on Pro · answered by openai/gpt-6-sol via OpenAI ·

  15. DeepSeek V4 Pro

    Passed: Graded 4.0 of 5 on average (lowest 3).

    Learning to Cook My Mother’s Recipes After She Moved Away
    I stand before the stove where she once stood,
    Her written cards beside the salt and sage.
    The kitchen fills with onions, oil, and wood,
    A recipe she scrawled on a torn page.
    She moved away, but left her wooden spoon,
    The cast-iron pan still seasoned black and deep.
    Her voice comes through the phone each afternoon:
    “Add stock, not water; let the garlic sleep.”
    I stir the pot the way she stirred for years,
    And measure by the hollows of my palm.
    The steam becomes a bridge across the tears,
    And in the scent, my mother’s kitchen calm.
    Though she is gone to some far other town,
    Her recipes still cook her presence down.

    360 tokens in, 4,656 out (4,498 of them reasoning) · 34.4 s · $0.0187 · 1 message on Pro · answered by deepseek/deepseek-v4-pro-0813 via Wafer ·

  16. GPT-6 Astra

    Passed: Graded 4.7 of 5 on average (lowest 4).

    Learning My Mother’s Recipes
    
    Since Mother moved away, I learn to make
    The meals she used to set before my chair.
    I dust the counter white with my mistake,
    And find her penciled notes still waiting there.
    
    I knead the dough; my hands are clumsy, slow.
    She wrote, Add water till the dough feels right.
    The page leaves out the things her fingers know.
    I hold it to the fading kitchen light.
    
    I set a bowl beside her empty plate;
    I call, and hear her kettle start to roar.
    She tells me bread is worth the time we wait;
    I listen as she shuts a cupboard door.
    
    I break the bread; she listens on the line.
    The loaf is hers; the crooked cuts are mine.

    367 tokens in, 1,191 out (1,034 of them reasoning) · 122.7 s · $0.0316 · 1 message on Pro · answered by openai/gpt-6-astra via OpenAI ·

  17. Claude Opus 5.5

    Passed: Graded 5.0 of 5 on average (lowest 5).

    Her Handwriting on the Index Cards
    
    The cards she left are stained with fat and flour,
    her looping script gone faint where butter spread;
    "A pinch," she writes, "then simmer half an hour,"
    as if I knew the rest of what she said.
    She moved away before I learned the flame,
    and now her kitchen lives inside my own:
    the onions sweat, I burn them all the same,
    the garlic sticks, the rice turns hard as stone.
    I call her with the phone against my ear,
    the pot lid rattling while she talks me through;
    "Now lower it," she says, "you're almost here,"
    and steam fogs glass the way it used to do.
    Tonight the stew tastes nearly as she planned;
    I eat it slow, and start to understand.

    498 tokens in, 1,541 out (1,297 of them reasoning) · 17.7 s · $0.0328 · 1 message on Pro · answered by anthropic/claude-opus-5.5 via Google ·

  18. Grok 4.7

    Passed: Graded 4.0 of 5 on average (lowest 3).

    Learning My Mother's Recipes
    
    Since Mother moved away, I cook her food
    from cards she wrote in pencil, faint and true.
    The onions weep the way they always would
    when Mother stood beside me, stirring stew.
    
    I turn the page and find her looping hand
    above a list of spices, salt, and bay.
    She moved away; I learn to understand
    each note still carries what she used to say.
    
    The spoon goes round; the broth begins to clear,
    and something in the steam recalls her face.
    I taste, and taste again, and persevere,
    until the missing kitchen finds its place.
    
    The evening settles; supper, warm and near,
    becomes my mother's recipes, held here.

    1,601 tokens in, 9,250 out (9,107 of them reasoning) · 87.9 s · $0.0570 · 1 message on Pro · answered by x-ai/grok-4.7 via xAI ·

  19. Claude Fable 5.1

    Passed: Graded 5.0 of 5 on average (lowest 5).

    The Cards She Left
    
    The cards she left are soft with thirty years,
    the pencil blurred where butter spat and dried;
    I read them at the stove, and no one hears
    me ask how much a pinch is, or how wide.
    She wrote "a handful," meaning hers, not mine,
    and "simmer till it smells the way it should";
    I call her, now three time zones down the line,
    and tilt the phone to show the pot. "That's good,"
    she says, "but let the garlic go a bit
    more gold." I do. The kitchen fills with her,
    the steam, the salt, the cardamom, and it
    is nearly right, the way old photos blur.
    I cook it wrong enough to need her still,
    and set two plates, and let the second chill.

    498 tokens in, 1,611 out (1,376 of them reasoning) · 24.1 s · $0.0855 · 1 message on Pro · answered by anthropic/claude-fable-5.1 via Google ·

More AI poem generator test inputs

The other AI poem generator test inputs, each with every model's reply, and the results across all five.

Questions

Which AI does best on “A sonnet”?

13 of 19 models passed this AI poem generator test input's check (rubric); Claude Sonnet 5.5, Claude Sonnet 5, Claude Haiku 4.5, Gemini 3.1 Pro, GLM 5.3, and Mistral Large 4 didn't. The cheapest reply that passed was GLM 5.3 Flash's, at $0.00014, and the fastest too, in 5.5 s. The dearest reply, Claude Fable 5.1's, cost 594 times as much ($0.0855).

What does a reply to “A sonnet” cost?

Through the models' APIs, what OpenRouter charged us ran from $0.00014 (GLM 5.3 Flash) to $0.0855 (Claude Fable 5.1) for this test input. In llmwise you don't pay by the token: a reply like these counts as one message on Pro, whichever model answers.

Claude, GPT, Gemini, DeepSeek, Grok, Kimi, GLM, and Mistral, 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.