Skill marketplace
Ready-made research methods, from deep reading to tracking new work.
Ask a question. Parallax reads hundreds of arXiv papers and writes a fully cited review.Pick your topics, and new papers arrive screened every day.
Content is AI-generated; verify against the paper.
Used by Researchers at
Search by keyword and by meaning in parallel, check every candidate, and sort the results by research direction. Build the search as a node graph, and add or change routes as you go.
Powered by Jev, a decision model, Parallax screens 10,000 papers in about 2 minutes, 10× faster and 5× cheaper than a general LLM, widening what one study can reach.
Run keyword and semantic searches in parallel to gather thousands of candidates at once, catching related work no matter how it's phrased.
Mixture-of-Recursions: Dynamic Recursive Depths
Every inclusion comes with its reasons: each paper's probabilities, and the criteria behind every screening and classification step, are recorded in the set, ready to check.
Tools, skills and context all sit within the agents' reach. Review any step, or take over whenever you want.
Ready-made research methods, from deep reading to tracking new work.
Collections, papers, sets, notes and reports all come with tools agents can use directly.
New papers from every arXiv field, from physics and mathematics to computer science, biology and economics, synced each day.
Ask questions, read full-text translations and get research reports in your language, always cited to the original.
File papers by topic and mark what you've read, with each collection's progress at a glance.
Agents read through the papers, track every source and write it up as a survey, a related-work section, a field review or a daily digest, however long the work takes.
Surveys
A survey of 478 papers on on-policy self-distillation, where a language model learns from itself given privileged context. It covers what the teacher sees, signal shaping, teacher dynamics, coupling with RL, the debates over when OPSD helps and what privileged information adds, and applications from agents to diffusion models.
Surveys
Test-time training lets a model keep learning after deployment by optimizing an explicit objective on the test input, context, problem or stream. This survey of 605 papers organizes the field by what is learned at test time and from which signal — self-supervised TTT, TTT layers and test-time memory, LLM test-time training, test-time RL, and agents and robots — writes them in one notation, and weighs five open debates.
Surveys
A looped (or recurrent-depth) transformer applies the same weight-tied block repeatedly to a token's hidden state, turning "depth" from a fixed property of the parameter count into a variable that can be spent at inference time.
Surveys
The capability of an agent increasingly depends not on its base model but on the harness built around it: instructions, context management, memory, tools and skills, control flow and orchestration, and the agent's own code.
Digests
October 2026 turned world-model research from building to measuring: controlled ablations and latency budgets for world action models, memory designs that keep interactive video worlds consistent on revisit, and evaluations showing that strong reconstruction and success-rate agreement can coexist with latents that miss physics and simulated failures in the wrong place.
Digests
August 2026 was the month world action models (WAMs) stopped being a slogan and turned into an engineering agenda. The shared template that emerged across the screened corpus is a video- or latent-dynamics backbone that predicts scene evolution together with an action expert…
Digests
August 2026's multimodal benchmark literature is unusually convergent. Rather than announcing new tasks, it rebuilds evaluation around the specific step where current models break: an unanswerable clip a model will not flag, a second image it cannot keep straight, a polygon it…
Digests
September 2026 moved video generation forward on seven fronts at once, and the three changes worth remembering are not about image quality. Agentic orchestration became an object of study rather than a wrapper: two systems now train the controller that composes requests, one…
Research Methods
Most guides give the search step one sentence and the writing step one slogan. This one covers both: how to know you've found everything, and how to turn a pile of papers into an argument.
Product Updates
General-purpose agents search the literature through web search APIs, which return only the top-k results and a lot of noise, so they can’t be exhaustive. Parallax runs full-corpus recall over about 1.58 million arXiv papers across all disciplines since 2020, and models retrieval as composable operations on sets of papers. Keyword search is sped up with an adjacent-word fragment index. Semantic search uses halfvec ANN pruned by year and discipline, with a provably lossless exact batching path as a fallback. Screening and classification use Jev’s calibrated probabilities, over 10x faster than our LLM-based pipeline at less than a fifth of the cost. This post covers the system’s design and the performance work needed to run it in production.
Can't find an answer? Email us
Parallax is an AI literature review tool built on arXiv. Its agent searches, screens and classifies papers directly, and every citation in an answer links to the source passage. Search history, collections and reports stay in your workspace for next time, so you don't have to shuttle context between chats, PDFs and notes.
Yes. Describe your question and scope, and the agent searches arXiv by keyword and by meaning, screens the candidates, reads the full texts in parallel and writes a survey, a related-work section or a field review, with every claim cited to its source passage. You can edit the result and export it as Markdown; check key conclusions against the original papers before you rely on them.
Yes. Schedule an agent with your topics and criteria, and it runs daily, weekly or monthly: each run goes through the new arXiv papers in your fields, keeps the relevant ones and delivers a digest with short summaries and the papers worth a close read. Open any run to ask follow-up questions.
We cover papers in every arXiv field, including physics, mathematics, computer science, quantitative biology, statistics and economics, updated since 2020. Older papers and sources beyond arXiv are on the way. Full-text access depends on licensing and processing status; when it isn't available yet, you can read the abstract and reading notes, or open the original source.
Click a citation in an answer to jump to the paper with the source passage highlighted, then check it in context. AI can still misread material or overreach; a citation is a route to verification, not a guarantee of correctness, so check key conclusions against the original experiments and conditions.
Yes. The free plan is free forever, with no card required. You don't need to learn prompting first either: enter a research question or paste an arXiv link and say what you want to understand. See the pricing page to compare plan limits.
No. When it needs background, the agent can look up your earlier chats, notes, collections and reports, and you can point it to a specific one. If you've just changed your goals or key constraints, mention them again in the new chat.
No. Your reports, notes and chats are not used to train our own models. They stay in your account and are never made public or shared with others without your consent.
Yes. Reports generated for you belong to you; you can edit them, export them as Markdown and publish them. Two caveats: purely AI-generated content may not be protected by copyright in many jurisdictions, and rights in the papers you cite remain with their authors, so cite your sources and follow any AI-disclosure rules set by journals, conferences or your institution. See section 6.4 of the Terms of Service.