Tested prompt · AI poem generator
A poem for a friend's marathon: every AI model's reply, tested
We ran this everyday 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
15 of 19 models passed this AI poem generator test input's check (key facts); Claude Fable 5.1, Claude Sonnet 5, Kimi K3, and GLM 5.3 didn't. The cheapest reply that passed was GPT-6 Luna's, at $0.000050; the fastest, DeepSeek V4.1 Flash's in 1.4 s. The dearest reply, Claude Fable 5.1's, cost 468 times as much ($0.0234).
The prompt, as sent, and its check
Checked by key facts, the same way for every model.
A poem for a friend's marathon (everyday)
Write an original poem about this, with a title on the first line. Keep every name and detail in it. --- For my friend Tomasz, who just ran his first marathon in Warsaw.
Sent with the AI poem generator's own instructions as the system prompt, as a free run of the tool sends them.
The answer must state “tomasz”, “marathon”, “warsaw”.
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.
| Model | Result | Cost | Time | Reply |
|---|---|---|---|---|
| Claude Fable 5.1Anthropic | Failed: missing "marathon". | $0.0234 | 8.7 s | 369 tokens |
| Claude Opus 5.5Anthropic | Passed: Stated all 3 facts with the section. | $0.0167 | 9.4 s | 296 tokens |
| Claude Sonnet 5.5Anthropic | Passed: Stated all 3 facts with the section. | $0.0038 | 4.2 s | 280 tokens |
| Claude Sonnet 5Anthropic | Failed: missing "marathon". | $0.0046 | 6.2 s | 328 tokens |
| Claude Haiku 5.5Anthropic | Passed: Stated all 3 facts with the section. | $0.00019 | 2.7 s | 285 tokens |
| Claude Haiku 4.5Anthropic | Passed: Stated all 3 facts with the section. | $0.0012 | 2.8 s | 163 tokens |
| GPT-6 AstraOpenAI | Passed: Stated all 3 facts with the section. | $0.0049 | 19.3 s | 125 tokens |
| GPT-6.1 SolOpenAI | Passed: Stated all 3 facts with the section. | $0.0014 | 5.2 s | 119 tokens |
| GPT-6 SolOpenAI | Passed: Stated all 3 facts with the section. | $0.0010 | 2.7 s | 91 tokens |
| GPT-6 LunaOpenAI | Passed: Stated all 3 facts with the section. | $0.000050 | 2.0 s | 97 tokens |
| Gemini 3.1 Pro (preview)Google | Passed: Stated all 3 facts with the section. | $0.0064 | 8.8 s | 141 tokens |
| Gemini 3.8 FlashGoogle | Passed: Stated all 3 facts with the section. | $0.00044 | 2.5 s | 161 tokens |
| DeepSeek V4.1 FlashDeepSeek | Passed: Stated all 3 facts with the section. | $0.00046 | 1.4 s | 144 tokens |
| DeepSeek V4 ProDeepSeek | Passed: Stated all 3 facts with the section. | $0.0014 | 3.0 s | 158 tokens |
| Grok 4.7xAI | Passed: Stated all 3 facts with the section. | $0.0032 | 6.3 s | 140 tokens |
| Kimi K3Moonshot | Failed: missing "marathon". | $0.0022 | 8.0 s | 172 tokens |
| GLM 5.3Z.ai | Failed: missing "marathon". | $0.0011 | 1.3 s | 144 tokens |
| GLM 5.3 FlashZ.ai | Passed: Stated all 3 facts with the section. | $0.00014 | 2.4 s | 154 tokens |
| Mistral Large 4Mistral | Passed: Stated all 3 facts with the section. | $0.0056 | 22.9 s | 83 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 4 replies that failed first, then the ones that passed, from the cheapest up.
GLM 5.3
Failed: missing "marathon".
First Miles in Warsaw The gun sounds on the Wisła morning, and Tomasz runs his first long race— past the old town's rebuilt walls, past runners in their thousands, each breath a small flag raised. Twenty-six miles of pavement, of blistered faith, of legs that argue and a heart that keeps replying no, further. The Vistula glitters; somewhere a church bell rings out over cobblestones that survived what the century could throw at them. At the finish line he doubles over, medal swinging against his chest, sweat and October light on his face— the whole of Warsaw, it seems, running with him.
373 tokens in, 160 out (16 of them reasoning) · 1.3 s · $0.0011 · 1 message on Pro · answered by z-ai/glm-5.3 via Friendli ·
Kimi K3
Failed: missing "marathon".
Tomasz Runs Warsaw At dawn the streets of Warsaw hold their breath, and Tomasz laces up beside the Vistula, one more runner in a river of runners, his heart a drum he has carried for months. Past the old town's rebuilt amber walls, past café windows and strangers calling his name, the kilometers stack like stones in his legs, and still he moves, and still the road gives way. Somewhere near thirty the doubt arrives, gray as November rain on the pavement, but Tomasz has trained for this exact weather, the storm that only lives inside the body. Then the finish line, the medal, the shaking hands, Warsaw opening around him like a cheer. My friend, you ran the whole impossible distance. Today the city knows your stride by heart.
452 tokens in, 199 out (27 of them reasoning) · 8.0 s · $0.0022 · 1 message on Pro · answered by moonshotai/kimi-k3 via Wafer ·
Claude Sonnet 5
Failed: missing "marathon".
For Tomasz, Crossing the Line in Warsaw Somewhere past the twentieth kilometer your legs stopped asking permission and just kept moving, the way the Vistula keeps moving, not fast, just faithful. Warsaw opened its old streets for you, Nowy Świat, the bridges, the grey stone watching another pair of shoes go by, and you were one more story it would keep. Tomasz, I know your lungs were fire somewhere near the last long stretch, that your knees sent up their small rebellions, and you told them no, not yet, not here. Then the line. The paper crackle of a medal. Someone's hand on your shoulder, maybe a stranger's. Twenty-six miles folded into your legs forever, and the whole city, quietly, applauding. First marathons don't end at the finish line— they end months later, when you're tying your shoes for no reason at all, and your body remembers Warsaw, remembers that it can.
493 tokens in, 357 out (29 of them reasoning) · 6.2 s · $0.0046 · 1 message on Pro · answered by anthropic/claude-sonnet-5 via Anthropic ·
Claude Fable 5.1
Failed: missing "marathon".
The Last Kilometre, Warsaw Tomasz, at the start you were only a number pinned to a shirt, a breath held in with ten thousand others, the Vistula flat and grey beside the bridges. By the tenth kilometre the crowd had thinned to strangers clapping mittened hands on Świętokrzyska, and your legs were still your own. Somewhere past the halfway mark the body stops asking why and simply argues: the calf, the hip, the small hot coin of pain beneath the left heel that you carried like a secret. You did not stop. Not at the water tables slick with cups, not on the long grey stretch where the wind came down the avenue and the old city towers refused to get any closer. Then the last kilometre, Tomasz, the one you told me you feared, and the barriers filled with faces, and the clock above the arch counting forward, forward, into a time that is now yours. Forty-two thousand one hundred and ninety-five metres of Warsaw, and at the end of it a foil blanket, a medal warm from your chest, and the ordinary miracle of standing still.
495 tokens in, 369 out · 8.7 s · $0.0234 · 1 message on Pro · answered by anthropic/claude-fable-5.1 via Google ·
GPT-6 Luna
Passed: Stated all 3 facts with the section.
The First Marathon For Tomasz, Warsaw opened mile by mile, morning light caught on the river’s slow ribbon. The streets unrolled beneath his steady feet, past windows bright with waking faces, past corners where the city seemed to cheer. His breath found rhythm; his stride held on, through every ache, each final stretch of road. Then the finish line rose, real beneath the sky— his first marathon, and Warsaw all around him.
364 tokens in, 127 out (30 of them reasoning) · 2.0 s · $0.0000 · 1 message on Pro · answered by openai/gpt-6-luna via OpenAI ·
GLM 5.3 Flash
Passed: Stated all 3 facts with the section.
Tomasz in Warsaw Forty-two kilometres of October morning, and Tomasz in his borrowed running shoes, breath fogging over the Vistula, the old city rising pale beyond the bridges. His legs learned a language of repetition— foot, breath, foot—while the crowds called his name in a language he knew by heart, their hands out like open doors. At kilometre thirty the wall came, and he went through it the only way he could: one stride, then another, then another, the finish line a red thread pulling him home. Now his medal weighs almost nothing, and everything—first marathon, first time his own two feet carried him further than he ever believed they could.
373 tokens in, 160 out (6 of them reasoning) · 2.4 s · $0.0001 · 1 message on Pro · answered by z-ai/glm-5.3-flash via Friendli ·
Claude Haiku 5.5
Passed: Stated all 3 facts with the section.
The Last Kilometre in Warsaw For my friend Tomasz, who ran his first marathon in Warsaw. Somewhere past the twentieth mile the city tilts, and the cobblestones begin to hum like wire. Your shoes are slabs of wet cement, your breath a torn flag snapping over the river. Strangers hold out oranges and paper cups. A drummer on a corner keeps the beat of a heart that will not be told to quit, and the crowd is one long, unbroken cheer. Then the finish arch, a gate of light, and you stop, hands on knees, laughing and crying, the medal cold against your chest, proof that the body can carry more than it thought. Tomasz, you ran your first marathon in Warsaw. Keep that sentence. Say it to yourself on the days when the road feels long again.
495 tokens in, 285 out · 2.7 s · $0.0002 · 1 message on Pro · answered by anthropic/claude-haiku-5.5 via Anthropic ·
Gemini 3.8 Flash
Passed: Stated all 3 facts with the section.
Across the Bridges of Warsaw The autumn chill hung thin above the Vistula when the starting horn broke the morning quiet. Tomasz, your shoes took the rhythm of the asphalt, mile after mile tracing the broad avenues. Past the red brick of the Old Town, under the shadow of the tall bridges, your breath turned steady, carving space through the heavy ache in your legs. The crowd blurred into a wall of clapping hands, the road rising toward the final bend. Twenty-six miles carried in the sinew, and forty-two kilometers left behind you. At the finish line, the city exhaled, your first marathon conquered in Warsaw, the medal cold against your chest, your stride written permanently into the stone.
370 tokens in, 161 out · 2.5 s · $0.0004 · 1 message on Pro · answered by google/gemini-3.8-flash via Google AI Studio ·
DeepSeek V4.1 Flash
Passed: Stated all 3 facts with the section.
For Tomasz, After His First Marathon in Warsaw At thirty kilometres the city turned to weather, the Vistula a grey smear past the embankment, and your legs, which had trusted you since Praga, began to send their small, polite refusals. You kept them. Past the Old Town's rebuilt walls, past the pigeons lifting off like thrown confetti, past strangers holding signs for someone else and shouting anyway, because you were running. Then the line — a thin white ribbon of chalk and tape, your name mispronounced over a loudspeaker, and forty-two kilometres folding shut behind you like a long street at dusk, going quiet, going yours.
382 tokens in, 290 out (146 of them reasoning) · 1.4 s · $0.0005 · 1 message on Pro · answered by deepseek/deepseek-v4.1-flash via Together ·
GPT-6 Sol
Passed: Stated all 3 facts with the section.
For Tomasz, in Warsaw My friend Tomasz crosses a painted line with his first marathon in his legs. Warsaw stretches behind him: pavement, crossings, windows catching light. His shoes stop moving, but his breath takes another moment to slow. Someone holds out water. Tomasz cups the paper cup in both hands, then lifts it to his mouth as the city goes on around him.
364 tokens in, 137 out (46 of them reasoning) · 2.7 s · $0.0010 · 1 message on Pro · answered by openai/gpt-6-sol via OpenAI ·
Claude Haiku 4.5
Passed: Stated all 3 facts with the section.
Tomasz Crosses the Line The streets of Warsaw held you up today, forty-two kilometers of your own becoming, your feet remembering each cobblestone and boulevard, each breath a small rebellion against the distance that tried to stop you. Nobody stays the same after running through their own city like that— you learned your limits were suggestions, that your body speaks a language older than the streets themselves. Somewhere near the finish line, the whole city was cheering for you, though you could only hear your heart, that faithful drum that carried you through every kilometer, every doubt. Welcome to the other side, Tomasz. You are a marathoner now, and Warsaw knows your name.
383 tokens in, 163 out · 2.8 s · $0.0012 · 1 message on Pro · answered by anthropic/claude-haiku-4.5 via Azure ·
GPT-6.1 Sol
Passed: Stated all 3 facts with the section.
Tomasz’s First Marathon in Warsaw My friend Tomasz, you have just run your first marathon in Warsaw. The city unrolled beneath your shoes: crosswalk stripes, asphalt, a scatter of cups, while each kilometre asked for something you were not yet sure you had. Then the finish line—a strip of ordinary ground made extraordinary by reaching it. Beyond it, you could stop. Your shoes could rest beneath a chair. But somewhere in Warsaw, the road still holds the shape of the first time you did not give up.
364 tokens in, 202 out (83 of them reasoning) · 5.2 s · $0.0014 · 1 message on Pro · answered by openai/gpt-6.1-sol via OpenAI ·
DeepSeek V4 Pro
Passed: Stated all 3 facts with the section.
Tomasz Runs Warsaw Sunday morning, the Vistula holding its breath, cobblestones older than any starting gun. You crossed the bridge at Praga, feet learning the river's long argument with stone. Kilometer thirty, the crowd a wall of voices calling your name like a small weather system. Tomasz, they shouted into your stride, and the city answered with its one endless avenue. At the finish, the medal's ribbon was a thin red canal around your neck, sweat from your first marathon still drying into the salt of Nowy Świat. Somewhere the clock stopped being a number and became the weight of your own shadow, finally still, finally yours, under the Warsaw sky that watched you run.
356 tokens in, 319 out (161 of them reasoning) · 3.0 s · $0.0014 · 1 message on Pro · answered by deepseek/deepseek-v4-pro-0813 via Wafer ·
Grok 4.7
Passed: Stated all 3 facts with the section.
First Finish, Warsaw Tomasz, the river kept pace beside you and would not look away. Forty-two kilometres of cobble and tram-hum, the Vistula flashing like a medal not yet pinned. Your first marathon, and the city opened its long grey throat for you. Old Town roofs the colour of cooled ash, bridges lifting their iron shoulders as your shoes struck the same stone twice. Somewhere past the halfway banners your breath became weather, small and stubborn, a private rain. Warsaw did not cheer in one voice. It clicked, it waited, it let you through. At the line your legs forgot their names and your hands remembered everything.
1,598 tokens in, 285 out (145 of them reasoning) · 6.3 s · $0.0032 · 1 message on Pro · answered by x-ai/grok-4.7 via xAI ·
Claude Sonnet 5.5
Passed: Stated all 3 facts with the section.
Tomasz at the Finish, Warsaw Before the start, your breath hung white above the crowd along the Vistula, and every shoelace was double-knotted against the mile you hadn't met. Then the long unspooling: bridges, cobbles, a stranger's cardboard sign, a child holding out orange slices like small suns on a paper plate. Somewhere past the thirtieth kilometre the body argued and you answered with one step, and then the next one, a plain and stubborn arithmetic. Now the medal swings against your chest, heavy as a bell that has finally rung, and your legs, those loyal, shaking engines, carry you through the noise to us. Tomasz, first marathon, first of many or the only one you'll ever need: either way, the city knows your name in the language of tired, happy feet.
495 tokens in, 280 out · 4.2 s · $0.0038 · 1 message on Pro · answered by anthropic/claude-sonnet-5.5 via Anthropic ·
GPT-6 Astra
Passed: Stated all 3 facts with the section.
The First Finish Tomasz, my friend, Warsaw unrolled beneath your shoes, a marathon measured in breath and pavement. At first, the road was something you could count; later, it was the next white stripe, the paper cup crushed lightly in your hand, the steady work of lifting one more knee. You have just finished your first marathon. Let those words settle beside your tired feet. The city keeps its buses, doors, and clocks, but you have crossed it at the pace of courage. My friend, the finish line is now behind you; the distance will always be yours.
364 tokens in, 125 out · 19.3 s · $0.0049 · 1 message on Pro · answered by openai/gpt-6-astra via OpenAI ·
Mistral Large 4
Passed: Stated all 3 facts with the section.
Tomasz's First Marathon Tomasz stands at the starting line in Warsaw, the Vistula glinting behind him, his shoes laced tight for forty-two kilometers. His friend watches from the crowd as the morning air turns cold. The gun fires. He runs past the Royal Castle, past the cobblestones of the Old Town, each step a first time. His legs burn at kilometer forty, but he does not stop. Arms raised at the finish line, sweat mixing with triumph, Tomasz has crossed into something new, his first marathon complete under the Warsaw sky.
368 tokens in, 2,570 out (2,487 of them reasoning) · 22.9 s · $0.0056 · 1 message on Pro · answered by mistralai/mistral-large-4-0 via Mistral ·
Gemini 3.1 Pro
Passed: Stated all 3 facts with the section.
The Distance in Warsaw The morning mist hung low over the Vistula when you pinned the paper numbers to your chest, Tomasz. Your very first marathon, forty-two kilometers stretching out across the concrete heart of Warsaw. You paced through the shadow of the Palace of Culture, sneakers striking a steady drumbeat on the wide avenue. We stood near the Old Town cobblestones and cheered as salt dried white on your forehead and shoulders. Every mile was a quiet battle in your lungs, but you kept pushing past the heavy ache in your calves. You crossed the painted line beneath the open sky, breathless, exhausted, and completely transformed.
370 tokens in, 1,009 out (868 of them reasoning) · 8.8 s · $0.0064 · 1 message on Pro · answered by google/gemini-3.1-pro-preview via Google AI Studio ·
Claude Opus 5.5
Passed: Stated all 3 facts with the section.
Tomasz, Kilometre Forty-Two Before dawn in Warsaw, Tomasz laced his shoes twice, pinned the paper number crooked over his heart, and joined the crowd breathing steam at the start line. The Vistula kept pace beside him, grey and patient, while strangers held out orange slices and paper cups he crushed and dropped like small applause. Somewhere past the halfway sign his legs turned to wet sand, and still he counted streetlamps, bridges, the next corner, the next, the next. This was his first marathon, the one no one could run for him, every stride paid for by months of dark mornings, by the friend who said you can and the doubt that said you won't. Then the finish arch, the city roaring its welcome, a foil blanket crackling on his shoulders like a flag, and a medal, cold and heavy, warming against his chest.
495 tokens in, 737 out (441 of them reasoning) · 9.4 s · $0.0167 · 1 message on Pro · answered by anthropic/claude-opus-5.5 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 poem for a friend's marathon”?
15 of 19 models passed this AI poem generator test input's check (key facts); Claude Fable 5.1, Claude Sonnet 5, Kimi K3, and GLM 5.3 didn't. The cheapest reply that passed was GPT-6 Luna's, at $0.000050; the fastest, DeepSeek V4.1 Flash's in 1.4 s. The dearest reply, Claude Fable 5.1's, cost 468 times as much ($0.0234).
What does a reply to “A poem for a friend's marathon” cost?
Through the models' APIs, what OpenRouter charged us ran from $0.000050 (GPT-6 Luna) to $0.0234 (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.