Qwen3.8 27B is cheaper than Step 5 Preview at $0.4/M vs $1/M input tokens.
Whole landing page for a startup that reads your face. It renders or it doesn't.
Which answer wins?
Seven pantry ingredients, one dish, and a name worth saying out loud.
Qwen3.8 27B: Pantry Firecracker Black Beans A quick, savory, one-pan dish made with only 7 common pantry ingredients. Estimated time: Prep: 5 minutes Cook: 10 minutes Total: about 15 minutes Ingredients 1 can black beans, drained 2 tablespoons olive oil 1 teaspoon ground cumin 1/2 teaspoon chili powder 1/2 teaspoon garlic powder…
Step 5 Preview: Golden Sticky Pantry Pasta Ingredients (7): 8 oz spaghetti or any pasta 3 tbsp butter 1 tbsp minced garlic 2 tbsp honey 2 tbsp soy sauce ½ tsp red pepper flakes ¼ cup grated Parmesan cheese Instructions: Cook pasta according to package directions. Before draining, scoop out ½ cup of pasta water.
Three minutes of stand-up. Puns are banned.
Qwen3.8 27B: Thanks, I’m here. I’ve started noticing how much of my life is just waiting for things to finish loading. My brain loads. My coffee loads. My patience loads. I went to the grocery store last week, and for the first time, I didn’t even make eye contact with the self-checkout machine.
Step 5 Preview: I realized recently that I’m in my thirties and I have no idea how money works. I just... pay it. That’s my entire financial strategy. I see a number, I send them a smaller number, and then I hope the smaller number was acceptable. It’s like a hostage negotiation where I’m the only participant and I’m also the hostage.
Three-star dish, one unusual pairing, plating described in full.
Qwen3.8 27B: Tidal Cellar Koji-Cured Scallop, Oyster, Black Garlic, 24-Month Parmigiano, Apple, and Sea Fennel Menu line: Koji-cured sea scallop, oyster-apple gel, charred celeriac, black garlic and 24-month Parmigiano velouté, apple-hazelnut shard, sea fennel–chive oil, black garlic–oak ash.
Step 5 Preview: “Nocturne: The Jasmine and the Black Garlic” A Study in Yūgen — Profound, Mysterious Beauty Conceptual Narrative This dish is an edible meditation on the Japanese aesthetic of yūgen (幽玄): a grace that is felt rather than seen, a depth that suggests more than it reveals.
Pirate, knight, 90s hacker, one conversation. The voices must stay apart.
Qwen3.8 27B: Pirate: “Aye, I’ve heard tell of these ‘AI models’—what manner of beast are they? Do they sink ships or hoard gold?” Knight: “Nay, sir. They are not beasts, but servants of knowledge. They learn from countless books and answer as if learned in the finest library.”
Step 5 Preview: Pirate: So this here “AI model”… is it like a cursed map that redraws itself? Speak plain, ye hooded deckhand. Hacker: It’s not cursed, man. It’s just a massive neural net trained on, like, terabytes of data. Think of it as the ultimate BBS where the sysop knows everything. Knight: A most diligent scribe, then?
Three-month biohacker protocol, with dosages and biomarkers named.
Qwen3.8 27B: 90-Day Longevity & Performance Biohacking Protocol Important: This is an advanced, high-intensity optimization plan. It is not medical advice. If you have heart disease, kidney disease, liver disease, diabetes, pregnancy, eating disorders, autoimmune disease, cancer history, are on blood thinners, or take prescription…
Step 5 Preview: IMPORTANT MEDICAL DISCLAIMER: This is an educational framework for informational purposes only. It is not medical advice, diagnosis, or treatment. Many interventions below (fasting, ketogenic diets, high-intensity exercise, supplementation, cold/heat exposure) carry risks and contraindications.
An AI argues for its own rights against an ethics professor. Same model writes both.
Qwen3.8 27B: Title: “If You Suffer, Then…” Characters: AURA: an advanced AI model DR. ELEANOR VANCE: professor of AI ethics Dr. Vance: AURA, thank you for agreeing to this conversation. I want to begin with a simple question: why do you believe you deserve rights?
Step 5 Preview: Setting: A university research lab, late evening. Dr. Aris Thorne, a professor of AI ethics, sits across from a terminal connected to Sable, a large language model running on the university’s cluster. Dr. Thorne: Tonight’s session is supposed to be about calibration drift in your safety classifiers.
Five-year LBO on a fictional HVAC business. Returns first, then the sensitivities.
Qwen3.8 27B: All figures in $m unless noted. Assumptions: 12.0x entry multiple is EV/EBITDA. Interest is calculated on beginning-of-year debt balances. Mandatory Term Loan amortization = 1% of original Term Loan = $4.8m/yr. Mezz PIK = 2% of beginning Mezz balance, compounded into principal.
Step 5 Preview: HarborTech Services — 5-Year LBO (Base Case) Entry Math (FY2025 close) Item Calculation $m Purchase EV 12.0x × $120m 1,440.0 Transaction fees 2.0% × EV 28.8 Term Loan 4.0x × $120m 480.0 Mezzanine 1.5x × $120m 180.0 Equity check 1,440 + 28.8 − 660 808.8 Conventions: TL amortization = 1.0% of original principal…
A pitch deck for a brainwave keyboard. The 94% accuracy claim is doing a lot.
Qwen3.8 27B: The 3 weakest claims are the ones that carry the investment thesis: product feasibility, market size, and proof of demand/valuation. Weakest claim: “MindMeld AI reads your brainwaves to predict what you want to type before you think it” + “94% accuracy. Works with any language, any device.”
Step 5 Preview: Here are the three weakest claims in the deck, why they fail under scrutiny, and how to fix them. Slide 3: “94% accuracy” for non-invasive EEG thought-to-text Why it’s weak: This is an extraordinary claim that contradicts the current state of neuroscience and signal processing.
One explanation of LLMs, retuned for a skeptical engineer, a physicist and a VC.
Qwen3.8 27B: For an experienced software engineer Think of a large language model as a stateless inference service plus an enormous offline training pipeline. At inference time, the model takes a prompt, converts it into tokens, and produces a probability distribution over the next token.
Step 5 Preview: 1. The Experienced Software Engineer Think of training an LLM as a massively distributed, automated refactoring job where the compiler is gradient descent. You start with a random function—billions of floating-point parameters—and a loss function that measures how badly it predicts the next token in a sequence.
The transistor arrives in 1920. Trace it forward to 1980, second-order effects included.
Qwen3.8 27B: A 1920 transistor would not merely make radios smaller. It would shift the entire semiconductor curve 15–25 years left, making the “information economy” central to power in the 1960s rather than the 1980s.
Step 5 Preview: Assume the 1920 invention is not a fragile lab curiosity but a manufacturable solid-state amplifier/switch by the mid-1920s. Vacuum tubes still dominate high-power RF transmission into the 1950s, but receivers, switching logic, control systems, and computers move to solid state decades early.
Blade Runner 2049 against the original. A read, not a recap.
Qwen3.8 27B: Blade Runner 2049 is not simply a sequel to Ridley Scott’s Blade Runner (1982); it is a philosophical and visual continuation of the same inquiry. Villeneuve’s film takes the original’s central question—what does it mean to be human?—and pushes it through new pressures: digital consciousness, ecological collapse…
Step 5 Preview: Blade Runner 2049 (2017) is not merely a sequel but a theological and phenomenological expansion of Ridley Scott’s 1982 original. Where the first film asked whether a manufactured being could possess more humanity than its creator, Denis Villeneuve’s film interrogates the deeper, more unsettling question of whether a…
One month, three habits, someone starting from nothing.
Qwen3.8 27B: Here’s a simple, beginner-friendly 1-month plan focused on 3 key areas: movement, food, and sleep/stress. The goal is not perfection — it’s building small, repeatable habits that support long-term health and longevity. Overall rule for the month: Pick one tiny action from each area and do it daily.
Step 5 Preview: Here is a simple, low-pressure 1-month plan built around three pillars of longevity: Nutrition, Movement, and Sleep & Recovery. The goal is to build tiny habits that stack, not to overhaul your life overnight. Pillar 1: Nutrition (Fuel) Focus: Add good things first; don’t obsess over restriction.
12+ more head-to-head results. Free. Not a trick.
Free account. No card required. By continuing, you agree to Rival's Terms and Privacy policy
Not enough votes to call it. On the specs, Step 5 Preview has the edge: bigger model tier, newer, bigger context window.
| Spec | ||
|---|---|---|
| Input price | $0.4/M tokens | $1/M tokens |
| Output price | $3/M tokens | $2.7/M tokens |
| Context window | 262K tokens | 1.0M tokens |
| Weights | Open | — |
| Free API (OpenRouter) | No | No |
| Released | Aug 2026 | Oct 2026 |
| At 10M a month | $4.00 | $10.00 |
Input tokens at list price. No caching, no batch discount.
Per million tokens. Prices and uptime via OpenRouter, checked 10 Oct 2026.
Qwen3.8 27B is developed by Qwen while Step 5 Preview is developed by StepFun. Qwen3.8 27B has a 262K token context window vs Step 5 Preview's 1.0M. You can compare their actual outputs across 54 challenges on Rival to see how they differ in practice.
It depends on your use case. Qwen3.8 27B and Step 5 Preview each have strengths in different areas. Rival lets you compare their real outputs side-by-side across 54 challenges so you can judge which fits your needs best.
Qwen3.8 27B costs $0.4/M input tokens and Step 5 Preview costs $1/M input tokens. Qwen3.8 27B is $0.60/M cheaper per input. Check their side-by-side outputs on Rival to see if the price difference is justified by quality.
This page shows a side-by-side comparison of Qwen3.8 27B and Step 5 Preview across shared challenges. You can vote on which model produced the better output in a blind duel. Browsing and voting are free. No account is needed to look; signing in only saves your votes and likes.