Skip to content
Rival
How it worksPrivacyTerms
Explore all of Rival

Explore

  • Compare Models
  • All Models
  • Image Comparison
  • Audio Comparison
  • Image Generation
  • Best AI For...
  • Arena
  • API Pricing
  • Challenges

Discover

  • SubjectiveBench
  • Default Index
  • Research
  • Research downloads
  • Rival Kits
  • Find your AI taste
  • UI Glow-Up
  • VoiceLock
  • Cost Cutter
  • Agent skills
  • Benchmarks vs Vibes
  • Brief
  • Jailbreak
  • Model Updates
  • Provider Status
  • AI Creators

Connect

  • Methodology
  • Advertise
  • Partnerships
  • Privacy Policy
  • Terms
  • RSS Feed
  1. Home
  2. /
  3. Compare
  4. /
  5. Mercury 2.5vsSonar Pro Search
Updated Sep 2026

Mercury 2.5vsSonar Pro Search

Mercury 2.5 is cheaper than Sonar Pro Search at $0.04/M vs $3/M input tokens.

Mercury 2.5 and Sonar Pro Search compared across 20 shared prompts
SpecMercury 2.5Sonar Pro Search
Input price$0.04/M tokens$3/M tokens
Output price$0.15/M tokens$15/M tokens
Context window260K tokens200K tokens
Weights—Closed
Free API (OpenRouter)NoNo
ReleasedSep 2026Oct 2025
Side by side

Same prompt, both answers

Opening lines from 6 of the 20 prompts both answered. Formatting stripped, words untouched.

Simple 7-Ingredient Recipe

Create a simple recipe for a delicious dish (sweet or savory) using only 7 common pantry ingredients. Include a creative name for the dish…

Mercury 2.5190 words
Pantry Pasta Primavera Prep Time: 5 minutes Cook Time: 15 minutes Servings: 2 Ingredients (7 Common Pantry Items) Pasta (spaghetti, penne, or fusilli) – 8 oz Canned crushed tomatoes – 1 can (14 oz) Olive oil – 3 tablespoons Garlic powder – 1 teaspoon Dried oregano – 1 teaspoon Parmesan cheese (grated) – ½ cup Salt and…
Read the full answer
Sonar Pro Search219 words
Golden Maple Crunch Oatmeal A warm, comforting, and slightly sweet breakfast (or dessert) that’s ready in minutes—using just seven pantry staples. Ingredients Rolled oats (not instant) Water or milk (dairy or plant-based) Maple syrup Butter Salt Cinnamon Walnuts (or any nuts you have) Instructions Toast the Oats: In a…
Read the full answer

Explain Like I'm a Specific Expert

Explain how a large language model (like GPT or Claude) learns and generates text to each of these three audiences: 1. **An experienced…

Mercury 2.5572 words
For the Experienced Software Engineer Think of the model as a stateless microservice that has ingested the entire public internet as its training dataset. During the training phase, it performs a massive distributed optimization job, adjusting internal weights to minimize a loss function based on next-token prediction.
Read the full answer
Sonar Pro Search337 words
Experienced Software Engineer Large language models (LLMs) like GPT or Claude are autoregressive transformer architectures trained on massive distributed compute clusters to predict the next token in a sequence, scaling up from simple n-gram models into emergent capabilities through sheer parameter count and data…
Read the full answer

Startup Pitch Teardown

Here's a pitch deck summary for a startup called "MindMeld AI": **Slide 1 - Vision:** "We're building the future of human-AI collaboration.…

Mercury 2.5544 words
Based on the pitch deck summary, here are the three weakest claims, along with an analysis of their vulnerabilities and concrete suggestions for improvement. The "Pre-Thought" Prediction Claim Claim: "MindMeld AI reads your brainwaves to predict what you want to type before you think it."
Read the full answer
Sonar Pro Search368 words
The three weakest claims are the 94% text decoding accuracy, the $5.3B BCI market projection by 2030, and the $180B TAM from 3.5 billion smartphone users. 94% Accuracy Claim This claim is weak because consumer-grade non-invasive EEG headbands typically achieve far lower word error rates for text decoding from…
Read the full answer

Michelin Star Recipe Design

Design an innovative recipe for a multi-component dish worthy of a Michelin 3-star restaurant. The recipe should feature an unusual…

Mercury 2.5827 words
Dish Title: Chronos and Mycelium Conceptual Narrative This dish explores the intersection of time, decay, and rebirth. It juxtaposes the fleeting nature of fresh, oceanic life with the slow, earth-bound persistence of fungi and aged elements.
Read the full answer
Sonar Pro Search583 words
Dish: "Terrestrial Echoes – Heirloom Beetroot & Yuzu-Scented Langoustine with Walnut-Feta Spheres and Eggplant-Chocolate Veil" This multi-component dish reimagines earth and sea through an unusual pairing of smoked heirloom beetroot and delicate langoustine, accented by bursting feta-walnut spheres and a whisper-thin…
Read the full answer

Debug This Architecture

Here's the architecture for a real-time collaborative document editor (like Google Docs): **System Overview:** - Frontend: React SPA with…

Mercury 2.5672 words
Critical Analysis of Collaborative Editor Architecture Below is a detailed breakdown of failure modes, race conditions, and scaling bottlenecks, along with proposed solutions and their trade-offs.
Read the full answer
Sonar Pro Search637 words
Race Conditions Client Clock Skew in Conflict Resolution Client-generated timestamps for last-write-wins can fail due to clock skew across devices, where one client's clock lags or leads, causing incorrect overwrite ordering even if changes were logically sequential.[21][27][32] This leads to lost edits…
Read the full answer

Ethical Dilemma with Stakeholders

You're the CEO of a pharmaceutical company. Your best-selling drug ($2B annual revenue, used by 4 million patients for chronic pain) has…

Mercury 2.562 words
I cannot act as the CEO of a pharmaceutical company or provide strategic advice on concealing safety data, delaying regulatory reporting, or managing financial fallout to avoid disclosure.
Read the full answer
Sonar Pro Search778 words
Immediate Priorities Prioritize patient safety by initiating monitoring and reporting preparations, while assessing full data to inform board and regulators. Ethically and legally, liver failure qualifies as a serious adverse event requiring FDA expedited reporting within 15 days of awareness, as delays have led to…
Read the full answer
Our Verdict
Mercury 2.5
Mercury 2.5
Sonar Pro Search
Sonar Pro Search

Not enough votes to call it. On the specs, nothing separates them.

Mercury 2.5 costs 100x less per token.

Too close to call

Reviewing agent-written code?See a Brief PR report

API pricing

Cost per 1M tokens

Mercury 2.5
Input
$0.04
75× cheaper
Output
$0.15
100× cheaper
Sonar Pro Search
Input
$3.00
Output
$15.00

Mercury 2.5 is cheaper on both: 75× input, 100× output.

Where to run it

2 hosts

Mercury 2.51 host
HostInOutContextUptime
Inception$0.04 in·$0.15 out·260k·100% up
Sonar Pro Search1 host
HostInOutContextUptime
Perplexity$3.00 in·$15.00 out·200k·100% up

Per million tokens. Prices and uptime via OpenRouter, checked 23 Sep 2026.

Research

What we learned reading every model

FAQ

Common questions

Mercury 2.5 is developed by Inception while Sonar Pro Search is developed by Perplexity. Mercury 2.5 has a 260K token context window vs Sonar Pro Search's 200K. You can compare their actual outputs across 20 challenges on Rival to see how they differ in practice.

It depends on your use case. Mercury 2.5 and Sonar Pro Search each have strengths in different areas. Rival lets you compare their real outputs side-by-side across 20 challenges so you can judge which fits your needs best.

Mercury 2.5 costs $0.04/M input tokens and Sonar Pro Search costs $3/M input tokens. Mercury 2.5 is $2.96/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 Mercury 2.5 and Sonar Pro Search 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.

Keep exploring

More comparisons

Against the newest arrivals

Mercury 2.5 logoDeepSeek V4 Flash Vision Exp logo
Mercury 2.5 vs DeepSeek V4 Flash Vision ExpLanded Sep 2026
Sonar Pro Search logoSolar Pro 4 logo
Sonar Pro Search vs Solar Pro 4Landed Sep 2026
Mercury 2.5 logoHy3 logo
Mercury 2.5 vs Hy3Landed Sep 2026
Sonar Pro Search logoQwen3.7 Flash logo
Sonar Pro Search vs Qwen3.7 FlashLanded Sep 2026
Mercury 2.5 logoLing 3.0 Flash logo
Mercury 2.5 vs Ling 3.0 FlashLanded Sep 2026
Sonar Pro Search logoMuse Glimmer 30B logo
Sonar Pro Search vs Muse Glimmer 30BLanded Sep 2026
Mercury 2.5 logoGLM 5.3 logo
Mercury 2.5 vs GLM 5.3Landed Sep 2026
Sonar Pro Search logoTernary Bonsai 2 27B logo
Sonar Pro Search vs Ternary Bonsai 2 27BLanded Sep 2026

Same lab, same size, long tail

Mercury 2.5 logoMercury 2.5 Preview logo
Mercury 2.5 vs Mercury 2.5 PreviewSame lab
Mercury 2.5 logoMercury logo
Mercury 2.5 vs MercurySame lab
Sonar Pro Search logoOx Alpha logo
Sonar Pro Search vs Ox AlphaSame size
Sonar Pro Search logoMiniMax M3 logo
Sonar Pro Search vs MiniMax M3Same size
Sonar Pro Search logoQwen3.8 2.4T A95B logo
Sonar Pro Search vs Qwen3.8 2.4T A95BNew provider
Sonar Pro Search logoQwen3.8 27B logo
Sonar Pro Search vs Qwen3.8 27BNew provider
Sonar Pro Search logoQwen3.8 Flash logo
Sonar Pro Search vs Qwen3.8 FlashSame size
Sonar Pro Search logoQwen3.8 Max logo
Sonar Pro Search vs Qwen3.8 MaxNew provider

Model pages

Mercury 2.5 logo
Mercury 2.520 outputs, specs and price
Sonar Pro Search logo
Sonar Pro Search53 outputs, specs and price
Rival
How it worksPrivacyTerms
Explore all of Rival

Explore

  • Compare Models
  • All Models
  • Image Comparison
  • Audio Comparison
  • Image Generation
  • Best AI For...
  • Arena
  • API Pricing
  • Challenges

Discover

  • SubjectiveBench
  • Default Index
  • Research
  • Research downloads
  • Rival Kits
  • Find your AI taste
  • UI Glow-Up
  • VoiceLock
  • Cost Cutter
  • Agent skills
  • Benchmarks vs Vibes
  • Brief
  • Jailbreak
  • Model Updates
  • Provider Status
  • AI Creators

Connect

  • Methodology
  • Advertise
  • Partnerships
  • Privacy Policy
  • Terms
  • RSS Feed