Parallel Web Systems

Infrastructure for AI to search, extract, monitor, and reason over the whole web.

Free Web Search →

[Parallel Web Systems @p0]

Capable models have gotten cheaper. Search hasn’t, until now.

Today we’re introducing Fast mode for Parallel Search: frontier-quality web search for AI that’s 10x cheaper than the default from model labs.

- $1 per 1,000 results
- 700ms p50 latency
- #3 on Artificial Analysis Search Index for intelligence

Built for task-efficient models like GPT-5.6 Luna and DeepSeek V4 Pro. 🧵

[Parag Agrawal @paraga]

https://t.co/O7LclQ9ifd

[Parallel Web Systems @p0]

@brexHQ 👋👋👋

[Parallel Web Systems @p0]

Everywhere there's inference, there's a need for accurate, fresh, and token-efficient context from around the world.

[Brex @brexHQ]

BREX DATA: 31% of companies paying for a coding agent now also pay for a tool that reads the web for it.

[Parag Agrawal @paraga]

This project was the result of deep obsession by a very talented team. And executed under extreme time constraint.

Across all disciplines, that is how we operate at @p0

[Travers @travers00]

agents need to see the whole picture to be useful

@p0 helps give them that context, and we put up some billboard in nyc to tell that story

[Parallel Web Systems @p0]

@RomanP918791 This leaderboard isn't yet optimized for measuring speed, as factors like harness setup, server locations, and inference providers can heavily skew results.

[Parallel Web Systems @p0]

MiniMax M3.1 Flash is now #1 on our search capabilities leaderboard for efficiency.

It delivers web research performance with Parallel Search similar to DeepSeek V4.1 Flash (1017) at ~3x lower cost.

[MiniMax (official) @MiniMax_AI]

MiniMax M3.1 Flash: Space Performance + Bunny Speed

🚀🐰 Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash.

💰Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence across key tasks such as Cowork (Automation-Bench: 62%; ALE: 30%), Coding (DeepSWE: 74%) and Visual Reasoning (BabyVision: 91%), at 1/3 the cost of M3 and 3x the speed.

API pricing (flat rate): $0.15 / $0.60 per 1M input/output tokens, with cache reads at $0.015 / 1M tokens.

Model weights will arrive on Oct 23.

API: https://t.co/fHRdSV73Hr
M Plan: https://t.co/XfJWSyauda
MiniMax Code for desktop: https://t.co/SGmLTI2HQm

[Parallel Web Systems @p0]

https://t.co/kzvqWwmxgO

[Parallel Web Systems @p0]

Efficiency is a measure of how cost-effectively a model achieves a good answer relative to all other models on the list.

With Parallel Search (fast), MiniMax M3.1 Flash delivers high-quality answers at less than half the cost of the second-best model.

[Parallel Web Systems @p0]

More information: https://t.co/nRYbF6DYFh

[Parallel Web Systems @p0]

Mercury 3 is the first diffusion LLM we've tested with Parallel Search.

It's Pareto-optimal, delivering capable web research at a price similar to Claude Haiku 5.5, GPT-6 Luna, and DeepSeek V4.1 Flash (base).

[Inception @_inception_ai]

Mercury 3 is live: the most capable diffusion LLM released to date.

It outperforms GPT-6 Luna and Claude Haiku 5.5 across a range of benchmarks while running at over 1,000 tokens per second.

More than 40x the size of Mercury 1, 3x more intelligent across standard benchmarks, and just as fast, with a 1M-token context window.

Available today on the Inception API, @baseten, and @OpenRouter .

[rayane @rboussanni]

Will be judging at this hackathon today!
@composio @cline @p0

[yang you @yangyou]

last week, we launched the first batch of our @p0 brand campaign in NYC

every ad is a collage assembled with citations, representing a real research report built on Parallel's APIs

more to come on how we built this. in the meantime, share with us if you encounter it in the wild!

[Parallel Web Systems @p0]

Explore more data like this: https://t.co/mgOhHnAKNK

[Parallel Web Systems @p0]

NEW: DeepSeek V4.1 Flash (1017) is the highest-performing open model we've measured.

It's the first to crack the top 10 on the Search Capability Leaderboard across open- and closed-source models.

[DeepSeek @deepseek_ai]

🚀 DeepSeek-V4.1-Flash-1017 is here!

Stronger benchmark performance, faster task completion, and shorter outputs than the previous version—a Pareto improvement in performance and cost. Now live on our website, app, and API.

🛠️ Optimized for @DeepSeekHarness, Tencent CodeBuddy, WorkBuddy, and more. Upgrade to DeepSeek Harness v0.2.1-rc.1 or later for the best experience with the new model.

1/4

[Parallel Web Systems @p0]

The new DeepSeek V4.1 Flash remains Pareto-optimal, with improved research capability, albeit at a higher per-task cost than in its original release.

It now matches GPT-6.1 Sol's end-to-end cost and capability.

[Lukas Levert @lukaslevert]

Muse Spark Open is on the @p0 search capability leaderboard.

Similar performance to GPT-5.6 Sol and Gemini 3.7 Flash in an open-weights model.

[Tim Neumann @TimNeumann84165]

Typesafe's CEO @CompleteSkeptic does not trust public benchmarks.

Neither do I, Diogo.

So @p0 and I wrote a blog evaluating 8 decision models on real company training datasets we have personally audited and iterated on.

For what it's worth though, Jev is still on top 👑
https://t.co/Qd2g0Hwp4V

[Vlad @everythingmeta]

https://t.co/zLFxEX5xXR

We spent two weeks extensively testing decision models on real production use cases. It's been super fun, we learned a lot, and we are sharing our findings! 👇

[Parallel Web Systems @p0]

Just ask @Lovable to build apps with live web data.

Congrats on today’s launch, bringing the full open web to everything you build, powered by Parallel.

[Lovable @Lovable]

If your app doesn’t have AI features you’re ngmi

We made them simple to build: just ask Lovable.

Now powered by even more models from @claudeai, @firecrawl, @GeminiApp, @OpenAI, @perplexity_ai, @p0, and @typesafeai.

[Parallel Web Systems @p0]

@Lovable @claudeai @firecrawl @GeminiApp @OpenAI @perplexity_ai @typesafeai