
founder/ceo @p0 https://nitter.cf/t.co/KLxULH5Io7, device following @vintweeta
San Francisco, CA
Joined April 2008
- Tweets3.7K
- Following1.5K
- Followers539K
- Likes4.8K
You should never have to compromise on accuracy, cost, and speed.
Parallel is #1 across the board on @ArtificialAnlys Search Index.
- Parallel Search Advanced (#1, score of 83)
- Parallel Search Fast (#4, score of 82)
And we promise to keep pushing the Pareto frontier for web data ☝️
Parallel Search scores 83 on the Artificial Analysis Search Index, level with Keenable Search (pro) at the top of the board, with a fast mode at 82 for $0.014 per task
We benchmarked two modes of @p0's Parallel Search API, fast and advanced, in our Stirrup agent harness with Parallel Search as its web_search tool to complete 1700 search-focused tasks.
In our August runs, advanced scored 75 and fast scored 73. In these November runs, advanced scores 83 and fast scores 82.
Key benchmarking results for Parallel Search:
➤ Artificial Analysis Search Index: Parallel Search (advanced, objective) scores 83, 1st of 26, and Parallel Search (fast, objective) 82, 4th. That is a ~50-point lift over the same model with no search (33). Advanced is level with Keenable Search (pro) at 83 and 3 points ahead of Exa (auto, dynamic highlights) and Perplexity (medium) at 80. Fast is slightly ahead of both
➤ DeepSearchQA is Parallel's strongest benchmark: both modes score 89 F1, 1st and 2nd of 26, comparable to Keenable Search (pro) at 88. Advanced also ranks 1st on AA-Omniscience (74%), just ahead of Keenable Search (pro) (73%), and 3rd on BrowseComp (87%, level with Keenable Search (pro) and Perplexity (medium))
➤ Low cost: fast costs about $0.014 per task, 2nd lowest of 26 behind TinyFish ($0.009), against a board median of $0.063. Advanced costs about $0.030 per task, 6th lowest
Key details:
➤ Setup: GPT-5.6 Luna (medium) in Stirrup, Parallel Search API with mode=fast and mode=advanced
➤ Tool shape: the model fills Parallel's objective (optional) and search_queries (1 to 3) fields on each call
➤ Pricing: $1 per 1k requests (fast) and $5 per 1k requests (advanced)
Parag Agrawal retweeted
Total Cost per Task combines search cost and model inference cost. Parallel Search (fast, objective) costs about $0.014 per task: about $0.005 for search (4.6 searches at $1 per 1k) and about $0.009 for the model.
Parallel Search (advanced, objective) costs about $0.030: about $0.023 for search (4.6 searches at $5 per 1k) and about $0.007 for the model.
On Artificial Analysis Search Index vs. Cost per Task, both modes are on the cost frontier: no product scores higher for less.
Parag Agrawal retweeted
parallel is dominating 3rd party and industry-specific benchmarks:
- #1 on AA search index + pareto optimal
- #1 on @ValsAI web search index
- #1 on range of benchmarks for finance, legal, healthcare/bio
You should never have to compromise on accuracy, cost, and speed.
Parallel is #1 across the board on @ArtificialAnlys Search Index.
- Parallel Search Advanced (#1, score of 83)
- Parallel Search Fast (#4, score of 82)
And we promise to keep pushing the Pareto frontier for web data ☝️
Parag Agrawal retweeted
Search from @p0 now completely dominates the pareto frontier on cost and quality on both @ValsAI and @ArtificialAnlys benchmarks. 🔥
Parallel is now the default web search for @Vercel
AI Gateway and @eve now have free web search from Parallel.
@p0 search is now native to AI Gateway and the default 𝚠𝚎𝚋_𝚜𝚎𝚊𝚛𝚌𝚑 for eve agents. Offer runs through January 2027.
vercel.com/changelog/paralle…
Announcing Parallel for Healthcare: upgrades to our index and ranking that put Parallel on the cost-accuracy frontier for healthcare AI search.
- Parallel delivers top accuracy across four healthcare benchmarks: PWS-Health, LitQA3, TrialQA, and MedBrowseComp
- On PWS-Health, our new healthcare-focused benchmark, Parallel delivers 1.4x more correct answers per dollar than the next-best engine
- On LitQA3, Parallel is the most accurate engine, with 1.7x more correct answers per dollar than the next-best engine
Healthcare answers sit deep in FDA labels, ClinicalTrials.gov records, and the methods sections of papers, rarely in the abstract.
@AcuityMD verified 574K records at 1/5 the cost of a frontier model using Parallel.
Gemini 4 Hydrogen is now widely available on
@googlecloud. We’re thrilled to be day-one launch partners.
Use Parallel to ground Gemini 4 Carbon natively and bring accurate, fast, and agent-ready web data to your production AI.
Get started: docs.cloud.google.com/gemini…
Fable 5.2 (@AnthropicAI) has reclaimed the #1 spot on Parallel's Search Capability Leaderboard, followed by Opus 5.5 and Hy4 (@TencentHunyuan), the highest-ranking open model we've tested.
When this leaderboard launched, Fable 5.1 led with a score of 73.4, which models like Sonnet 5.5 now beat through intelligent use of web search.
Announcing Parallel for Startups:
Free credits for teams at the frontier of the agentic web.
Available today to founders and teams backed by @ycombinator, @speedrun, @neo, @joinef, @fdotinc, or @nullfellows.
Not in a partner program? Apply at parallel.ai/startups?utm_sou…
Parag Agrawal retweeted
there's a lil' sidekick waiting for you on the Parallel for Startups webpage. his name's Clicky ꒰১🖱️໒꒱
Announcing Parallel for Startups:
Free credits for teams at the frontier of the agentic web.
Available today to founders and teams backed by @ycombinator, @speedrun, @neo, @joinef, @fdotinc, or @nullfellows.
Not in a partner program? Apply at parallel.ai/startups?utm_sou…
Parag Agrawal retweeted
come check out my talk about webjev at @aiDotEngineer !
We've got a big week ahead at @aiDotEngineer Code Summit in SF:
• Tonight @ 5:30 pm: the welcome reception, co-hosted with our friends at @AnthropicAI
• Tomorrow @ 6:30 pm: happy hour with @factory (link in comments to request a spot)
• Thursday, 2:20 pm: @devingupta_ presents "The first Grounded Decision Model" on WebJev, our new Grounded Decisions API that answers questions using evidence from the live web
• All week: stop by our booth to say hello
Parag Agrawal retweeted
parallel is hiring across every team
- research + mts (search ranking, model training, compute scaling)
- deployed engineers
- design (brand, product, design eng)
- gtm
- marketing
- finance, legal + ops
sf, palo alto, nyc, london. if you want to build the web for its second user, come join us / tag them below
Parag Agrawal retweeted
Just in time for the weekend, happy building
Parag Agrawal retweeted
sub-second grounded decisions over the live web,
completely free
with an openrouter api key
right now!
go try it out!
Parag Agrawal retweeted
What's better than the highest accuracy web-grounded decisions model? How about a week of free requests?
Try Parallel Decisions to get fast structured answers about the live web with probabilities and citations, free for 7 days exclusively for OpenRouter users!
Parallel Decisions is now on OpenRouter. Parallel Decisions delivers the highest accuracy on web-grounded decisions for System One models like Jev. Get web-grounded answers 13× faster at 1/16 the cost of OpenAI’s GPT Luna with native web search.
Ask questions about the live web and get structured answers with probabilities and citations in under a second.
Parallel Decisions is now on @OpenRouter
Free for the next 7 days.
It delivers the highest accuracy for web-grounded decisions, with answers in under a second. Get web-grounded answers 13× faster at 1/16 the cost of OpenAI’s GPT Luna with native web search.
Select Parallel: Decisions Fast in the model list, or set your model ID to parallel/decisions-fast. We can’t wait to see what you build.
Learn more: parallel.ai/blog/decisions-o…
Parallel Decisions is now on OpenRouter. Parallel Decisions delivers the highest accuracy on web-grounded decisions for System One models like Jev. Get web-grounded answers 13× faster at 1/16 the cost of OpenAI’s GPT Luna with native web search.
Ask questions about the live web and get structured answers with probabilities and citations in under a second.































