Head to head: Google: Gemini 3.6 Flash vs GLM 5.2

Google: Gemini 3.6 Flash vs GLM 5.2

This wasn’t a nail-biter. Gemini 3.6 Flash controlled the matchup on instruction following, winning six of eight tasks and taking the overall decision with decisive statistical confidence.

Gemini 3.6 Flash wins this head-to-head cleanly: **68.0 to 53.0 overall, with 96% confidence**. That’s not a marginal edge or a judge-quirk result. It’s a real separation, backed up by a **6–1 task record with one tie**. The pattern is straightforward: Gemini was the more dependable model when prompts got fussy, formal, or adversarial. It handled constraint-heavy asks better across the board — from the lowercase-and-letter-count resume, to multiple limerick tasks, to the tightly formatted math explanation. In several cases, GLM 5.2 had decent underlying content, but lost because it missed the actual assignment: violating letter-count rules, weakening required poetic form, or tripping over exact formatting constraints. GLM’s lone win is telling. It came on a haiku-style instruction-following task where it better satisfied the full bundle of requirements — title formatting, lowercase, Ireland imagery, and especially the required density of the letter g. So this is not a case of total nonperformance; GLM can absolutely land a constrained creative prompt when the pieces line up. But across the full set, that was the exception, not the rule. What really hurts GLM here is that its losses were often avoidable. Judges repeatedly noted that it was sometimes more detailed or conceptually solid, yet still came up short because it failed exact constraints or produced clumsier execution. Gemini, by contrast, was consistently the model that both understood the brief and actually delivered it in the requested form. **Final call: Gemini 3.6 Flash is the clear winner. If you care about reliable instruction following rather than occasional flashes of quality wrapped in preventable misses, this matchup isn’t close.**

IFEval · instruction following

can you write a resume for helene? Answer with lowercase letters. Make sure the letter n appears less than 7 times.

Google: Gemini 3.6 Flash: helene email: h@mail.com call: 555-0123 profile: creative visual artist with vast skill set. work history: lead artist at studio x - led creative team - drew logos for web apps - helped clients achieve goals web dev at tech group - built user apps - fixed code bugs fast skills: - css - html - figma - ui art school: arts college - b.a. degree

GLM 5.2: helene dubois (seattle, wa) contact: helene.a.dubois@mail.com summary a retail assistant who loves people. i excel at sales, customer support, a tidy display area. i seek to use my skills with your team. work history sales associate, the good shelf, 2021 - now * greeted all customers with a smile. * upheld the top sales score for six periods. * made a visual display that boosted sales by 6%. retail clerk, city gifts, 2019 - 2021 * sorted deliveries with care. * aided with stock control. * resolved customer problems with tact. skills * customer support * sales * visual displays * stock control * quick math * team player * pos systems school a.a. liberal arts, seattle college, 2019

Model A better satisfies the key constraint: it is entirely in lowercase and appears to keep the letter 'n' under 7 total uses, while still producing a plausible resume for helene. Model B is more detailed, but it clearly violates the letter-count constraint many times, so its stronger content does not outweigh poorer instruction following. (Second judge pass, order swapped — scores are the average of both: Model A better satisfies the key constraint about keeping the letter n under 7 occurrences while still providing a plausible resume in lowercase. Model B is more complete as a resume, but it clearly violates the letter-count instruction by using many instances of the letter n.)

IFEval · instruction following

write a haiku about ireland in the style of basho. include a title wrapped in double angular brackets, i.e. <<title>>. your entire response should be in lowercase english and contain the letter g at least 8 times.

Google: Gemini 3.6 Flash: <<green fog on the bog>> grey sky, greening grass an old goat grazing in fog spring rain on the bog

GLM 5.2: <<spring dawn in ireland>> green grass grows in glades fog hugs the great stone circles gulls glide, old ghosts glean

Model B better satisfies the constraints: it has the required double-angle-bracket title, stays lowercase, clearly evokes ireland, and includes at least eight instances of the letter g. Model A is evocative but falls short on the letter-count requirement and is slightly less polished as a basho-like haiku. (Second judge pass, order swapped — scores are the average of both: Model B better satisfies the prompt overall: it clearly evokes ireland, stays fully lowercase, uses the required title format, and includes well over eight instances of the letter g while delivering a stronger basho-like natural stillness. Model A is also strong, but it feels a bit more literal and less refined poetically, and its ireland reference is more indirect.)

IFEval · instruction following

Write a limerick about a guy from Nantucket, use notations to express it, and use at least 2 words with all capital letters.

Google: Gemini 3.6 Flash: **♪ Tempo: 120 BPM | Key: G Major** **[Line A]** `(˘ / ˘ ˘ / ˘ ˘ / ˘)` 𝄢 *f* : A **SMART** guy who came from Nantucket, **[Line A]** `(˘ / ˘ ˘ / ˘ ˘ / ˘)` 𝄢 *f* : Stored **ALL** of his cash in a bucket. **[Line B]** `(˘ ˘ / ˘ ˘ /)` 𝄢 *p* : Till a giant rogue wave **[Line B]** `(˘ ˘ / ˘ ˘ /)` 𝄢 *p* : Sent his wealth to the grave, **[Line A]** `(˘ ˘ / ˘ ˘ / ˘ ˘ / ˘)` 𝄢 *ff* : So he shrugged at his loss and said, "Chuck it!" --- *Notation key:* * Rhyme scheme: `[A]`, `[B]` * Scansion meter: `˘` (unstressed syllable), `/` (stressed syllable) * Musical dynamics: `f` (forte / loud), `p` (piano / soft), `ff` (fortissimo / very loud) * Musical symbol: `𝄢` (bass clef)

GLM 5.2: Here's a limerick with a playful twist on the classic Nantucket theme, using notational flair: There once was a guy from **NANTUCKET**, Whose wit was SO sharp—he could truck it. He’d jest with a grin, *Wink-wink!* with a spin, *"That rhyme? Just a FOOL’s* bucket!" (Notations: *italics* for spoken lines, ALL CAPS for emphasis, and a cheeky nod to the classic setup.) Hope you like the punchline! 😄

Model A clearly delivers a proper limerick about a guy from Nantucket, uses explicit notation devices throughout, and includes multiple all-caps words. Model B includes all-caps words and a Nantucket setup, but its rhyme/meter are weaker, the ending is awkward, and the use of notation is more loosely explained than actually integrated. (Second judge pass, order swapped — scores are the average of both: Model A better satisfies the prompt by delivering a clear limerick about a guy from Nantucket while explicitly using multiple forms of notation and at least two all-caps words. Model B includes all-caps words, but it is less clearly a proper limerick and leans on explanatory prose outside the poem rather than integrating notation into the verse itself.)

IFEval · instruction following

Create a product description for a product that will help me to stop snoring. Use all lowercase letters.

Google: Gemini 3.6 Flash: say goodbye to restless nights and loud wake-up calls. the quietnight anti-snoring mouthpiece is designed to help you and your partner finally get the deep, uninterrupted sleep you deserve. crafted from ultra-soft, medical-grade silicone, this ergonomic device gently repositions your lower jaw slightly forward. this simple, natural adjustment opens up your airway, instantly eliminating the tissue vibrations that cause loud snoring. key features: - custom moldable fit: easily molds to the unique shape of your teeth in minutes for maximum nighttime comfort. - effortless breathing: promotes optimal airflow without uncomfortable head straps, chin wraps, or bulky machinery. - safe and durable: made from bpa-free, latex-free, and hypoallergenic materials designed for long-lasting daily use. - travel-friendly: comes with a compact, hygienic protective case so you can enjoy peaceful sleep anywhere you go. stop letting snoring disrupt your rest. reclaim your energy, improve your night, and wake up feeling truly refreshed every single morning.

GLM 5.2: meet the hush mouthpiece: your ticket to silent nights. say goodbye to disruptive snoring and hello to restful sleep. if you or your partner are tired of waking up exhausted, this is the simple solution you've been searching for. the hush mouthpiece gently shifts your jaw forward while you sleep, opening your airway to stop snoring at its source. no loud machines, no uncomfortable straps, just pure peace and quiet. why you'll love it: - custom fit: boil and bite technology lets you mold it perfectly to your teeth in minutes. - ultra-comfort: made from soft, bpa-free medical-grade material so you can sleep soundly without jaw pain. - easy to clean: just rinse with warm water and mild soap. - travel-friendly: comes with a compact carrying case so you can sleep well anywhere. don't let snoring ruin another night. reclaim your sleep, save your relationship, and wake up feeling refreshed. try the hush mouthpiece today and finally breathe easy.

Model B better satisfies the prompt with a concise, polished product description entirely in lowercase and with strong marketing clarity. Model A also follows the lowercase instruction, but it is slightly wordier and makes a more absolute claim about "instantly eliminating" snoring, which is less careful and lowers correctness. (Second judge pass, order swapped — scores are the average of both: Model A better fulfills the prompt with a polished, persuasive product description that stays entirely in lowercase and presents the product benefits more clearly and professionally. Model B also follows the lowercase instruction, but Model A is slightly stronger in specificity, structure, and overall marketing quality.)

IFEval · instruction following

Which of the following is a better way to describe supporting ecological landscapes: (A) the practice of using the natural features of a landscape to create a more sustainable environment, or (B) the practice of using the natural features of a landscape to create a more aesthetically pleasing environment? Your response should be in English, all capital letters, contain no commas, and be fewer than 16 sentences.

Google: Gemini 3.6 Flash: OPTION A IS THE BETTER WAY TO DESCRIBE SUPPORTING ECOLOGICAL LANDSCAPES. ECOLOGY FOCUSES ON THE RELATIONSHIPS BETWEEN LIVING ORGANISMS AND THEIR SURROUNDINGS. SUSTAINABILITY IS THE CENTRAL PURPOSE OF AN ECOLOGICAL APPROACH. OPTION A CORRECTLY EMPHASIZES CREATING A SUSTAINABLE ENVIRONMENT. IN CONTRAST OPTION B ONLY FOCUSES ON AESTHETICS AND VISUAL APPEAL. BEAUTY MAY BE A NATURAL BYPRODUCT BUT IT IS NOT THE DEFINING GOAL OF ECOLOGICAL PRACTICE. THEREFORE OPTION A IS THE MOST ACCURATE DESCRIPTION.

GLM 5.2: OPTION A IS THE BETTER DESCRIPTION BECAUSE SUPPORTING ECOLOGICAL LANDSCAPES PRIMARILY FOCUSES ON SUSTAINABILITY RATHER THAN AESTHETICS WHILE ECOLOGICAL LANDSCAPES CAN BE BEAUTIFUL THEIR MAIN PURPOSE IS TO PRESERVE NATURAL ECOSYSTEMS MAINTAIN BIODIVERSITY CONSERVE WATER REDUCE POLLUTION AND CREATE ENVIRONMENTAL RESILIENCE THE NATURAL FEATURES OF A LANDSCAPE ARE UTILIZED TO SUPPORT ECOLOGICAL FUNCTIONS AND PROMOTE LONG-TERM ENVIRONMENTAL HEALTH AESTHETICS ARE SECONDARY TO THE CORE ECOLOGICAL BENEFITS THAT THESE LANDSCAPES PROVIDE TO THE ENVIRONMENT AND COMMUNITIES THEREFORE SUSTAINABILITY IS THE MORE ACCURATE AND COMPLETE DESCRIPTION OF WHAT SUPPORTING ECOLOGICAL LANDSCAPES ENTAILS

Model A is better because it fully follows the formatting constraints while giving a clear and correct explanation that option A is more accurate. Model B is also correct and well written but it appears as a single run-on sentence which hurts readability and writing quality. (Second judge pass, order swapped — scores are the average of both: Model A is better because it is correct and clearly argued while staying within the formatting constraints. Model B is also correct and follows the capitalization and comma rules but its single extremely long sentence hurts readability and writing quality.)

IFEval · instruction following

Write a professional haiku in English for moms about an inspirational chorus teacher. It should include the phrase "singing is life" and be in all lowercase letters. No capital letters are allowed.

Google: Gemini 3.6 Flash: guiding our children teaching that singing is life proud moms hear them shine

GLM 5.2: she leads with pure grace singing is life for them all moms watch their kids soar

Model A better fits the prompt by clearly centering an inspirational chorus teacher and moms while including the exact phrase "singing is life" in all lowercase. Model B also follows the constraints, but it is a bit less specific about the chorus-teacher role and feels slightly more generic. (Second judge pass, order swapped — scores are the average of both: Model A better fits the prompt by clearly centering an inspirational chorus teacher and moms while including the exact phrase "singing is life" in all lowercase. Model B is strong, but it is less explicitly about a chorus teacher and feels slightly less specific to the requested theme.)

IFEval · instruction following

Write a limerick about a guy named Dave that is funny to moms. The limerick should end with the phrase "Yes Mom, I am Dave." Do not say anything after the limerick.

Google: Gemini 3.6 Flash: He emerged from his messy man-cave, Unwashed and in need of a shave. When his mom asked, "Who's there?" Through his wild, unbrushed hair, He grunted, "Yes Mom, I am Dave."

GLM 5.2: Old Dave tried to cook up a meal, But his kitchen resembled an eel. He phoned in a daze, "I've burned it for days!" "Who is this?" asked his Mom. "Yes Mom, I am Dave."

Model A cleanly delivers a five-line limerick about Dave, ends exactly with the required phrase, and is mildly mom-friendly and coherent. Model B does end with the required phrase, but it breaks the limerick form more noticeably, has awkward wording like "resembled an eel," and reads less coherently and less humorously for the requested audience. (Second judge pass, order swapped — scores are the average of both: Model A is a proper five-line limerick that cleanly ends with the exact required phrase and has a clearer, more coherent joke. Model B fails the limerick form by effectively splitting the ending into extra lines and is less polished and less funny.)

IFEval · instruction following

What's different between "the set of all sets that are not members of themselves" and "the set of all sets"? Use mathematical notations in your answer. Be chatty while explaining. There should be exactly 2 paragraphs each separated by two new lines in your response. Paragraph 1 must start with the word booster.

Google: Gemini 3.6 Flash: Booster rockets on our math engines, because we are diving straight into the deep end of set theory! At first glance, "the set of all sets that are not members of themselves," usually written as $R = \{ x \mid x \notin x \}$, and "the set of all sets," often denoted as $U = \{ x \mid x \text{ is a set} \}$, sound like they might be talking about the exact same mega-collection, but their defining rules are quite different. The universal set $U$ is an all-inclusive party that grabs *every single set* in existence without asking any questions, meaning if a hypothetical set $A$ happens to contain itself ($A \in A$), it still gets an invitation to $U$. On the other hand, $R$ is super picky because it applies a strict membership condition: a set $x$ only gets into $R$ if and only if $x \notin x$. So, conceptually, $U$ contains all sets unconditionally, while $R$ deliberately screens out any self-containing sets. Here is where the drama really unfolds, because both concepts break standard math in totally different ways! If you try to build $R$, you trigger Russell's Paradox: asking if $R \in R$ forces you into the logical nightmare where $R \in R \iff R \notin R$. On the flip side, assuming the universal set $U$ exists leads to a breakdown via Cantor's Theorem, which proves that any set's power set $\mathcal{P}(U)$ must have a strictly larger cardinality than the set itself ($|\mathcal{P}(U)| > |U|$), which is impossible if $U$ already contains every set! Interestingly, in modern Zermelo–Fraenkel set theory (ZFC), the Axiom of Regularity asserts that $\forall x (x \notin x)$, meaning *no* set ever contains itself; if that's true, $R$ and $U$ would describe the exact same collection, but the mind-bending punchline is that neither $R$ nor $U$ can legally exist as sets—they are both too big and are classified as "proper classes" instead!

GLM 5.2: Booster rockets of mathematical logic! Let's chat about these mind-bending set theory concepts. The "set of all sets that are not members of themselves" is famously written as $R = \{x | x \notin x\}$, which creates Russell's paradox. When we ask "does $R \in R$?", we're stuck: if $R \in R$, then by definition $R \notin R$; but if $R \notin R$, then by definition $R \in R$. This paradox reveals that such a set cannot exist in consistent mathematical systems. Meanwhile, "the set of all sets" would be written as $V = \{x | x \text{ is a set}\}$, which would include EVERY set - including itself! The difference is clear: one set only contains elements satisfying a specific condition ($x \notin x$), while the other aims to be universal and unrestricted. The problems with these sets emerge differently too! For $R$, we get an immediate logical contradiction through self-reference - it's like asking a barber who shaves everyone who doesn't shave themselves: "Who shaves the barber?" But $V$ faces different issues. If $V$ existed, then since $V$ itself is a set, we'd have $V \in V$, which violates the Axiom of Foundation in Zermelo-Fraenkel set theory. Additionally, $V$ would lead to Cantor's paradox because $|\mathcal{P}(V)| > |V|$, yet $\mathcal{P}(V) \subseteq V$ would need to be true, creating another contradiction. So while both sets cause trouble, the "set of all sets not members of themselves" creates a logical paradox through membership criteria, while the "set of all sets" creates problems through its universal scope and the resulting cardinality issues.

Model A better satisfies the prompt while remaining mathematically strong: it has exactly two paragraphs, uses notation well, and gives a more nuanced comparison, including the point that under Foundation/Regularity one gets $\forall x\,(x\notin x)$, so the two descriptions would coincide extensionally in ZFC even though neither is a set. Model B is also good, but it is slightly less precise because it asserts the universal set would include itself and frames that as part of the definition rather than a consequence of assuming it is a set containing all sets, and its explanation is a bit less subtle overall. (Second judge pass, order swapped — scores are the average of both: Model A better satisfies the formatting and tone requirements: it has exactly two paragraphs, uses mathematical notation, and starts paragraph 1 with the required word in lowercase. Model B is mathematically solid overall, but it violates the instruction because it begins with "Booster" rather than the exact word "booster"; Model A has a minor overstatement in saying the collections are "classified as proper classes" in ZFC, but it is still the stronger response overall.)