Claude科研計劃擴圍

2026年8月28日 00:00
站內 AI 整理稿

Claude Team plan for scientistsPutting Claude in the hands of scientistsScientists at academic and nonprofit research institutions around the world can now get Claude at no cost to start.

Get startedGet startedGet startedMade for scientific researchA workbench built for scienceClaude generates and tests hypotheses, writes code, and runs analyses on your data.It scales compute from your laptop to hundreds of GPUs on demand, and traces steps from data wrangling to publication.

Grounded in scientific researchClaude cites its sources, shows the math behind results, and pushes back when the evidence doesn't support the claim.One plan that grows with youStart free, bring your group into one shared workspace, and scale up as your projects demand it.

Your research data stays yours.

ResearchHow Claude is accelerating protein design and analytical chemistryLearn moreLearn moreLearn moreHow scientists use ClaudeAstrophysicsFusion of astronomical datasetsCreate a map of the full sky in ultraviolet by combining and calibrating observations from different space telescopes.

Use statistical techniques to fill in regions with missing data.BiologyAnalyze biological data end to endAnalyze molecular datasets like scRNA, WGS, ATAC, MS-Spec, and more from raw feature counts to biologically resolved results.Claude scales your analysis and fits quantitative models on demand.

ChemistryPlan a synthesis route for any targetRun retrosynthesis on a drawn structure or SMILES, review each route for protecting-group strategy, selectivity, and step order, and land on a path from in-stock materials.

Quantum computingOptimize and verify quantum circuitsClaude rewrites fault tolerant quantum circuits and verifies changes, so work that used to take specialists months per circuit can run largely on its own.

StatisticsTest hypotheses computationallyRun numerical experiments, compare results against theoretical predictions, and generate publication-ready figures in a fraction of the time it used to take.

Materials scienceMaterials discovery and predictionWork backward from target properties to candidate structures, validate them with first-principles simulations, and screen for what can be synthesized.

Claude connects to your toolsWork across your literature databases, research platforms, and productivity tools in one place.

Explore connectorsExplore connectorsExplore connectorsProducts and programs to support scientistsClaude ScienceThe Claude Science app is your workbench for rigorous scientific research.

It runs analyses, searches databases, and traces steps from data wrangling to publication, so you can spend more time on science.Learn moreLearn moreLearn moreAI for ScienceAnthropic’s grant program for ambitious research funds months of sustained work through extra usage credits.

We’ve backed hundreds of projects so far across institutions and scientific domains.Learn moreLearn moreLearn morePutting Claude in the hands of more scientists10,000 scientists around the world can now get Claude at no cost to start.

Verified principal investigators qualify for a Claude Team subscription plan and then add their research team to Standard seats for free, or Premium seats for $15 per month.‍‍‍Promotional pricing for 12 months for verified research groups.Limited availability; terms apply.

Standard$0/monthregularly $20/monthVerify nowVerify nowVerify nowAccess to Claude Science, Code, & CoworkConnectors to scientific databases and skillsShared projects and file creationCentral billing and administrationSingle sign-on (SSO)Minimum 1 seatPremium$15/monthregularly $100/monthVerify nowVerify nowVerify nowEverything in Standard, plus:5x more usageHigher limits for long-running analysesExtra usageApply for credits‍Learn moreLearn moreLearn moreWhen a project outgrows existing capacity, apply for extra usage credits through the AI for Science program.

Extra usage limits apply.Prices shown don’t include applicable tax and may change at Anthropic’s discretion.1–25 seats per group; contact us for larger groups.Annual billing available ($0 / $180 per seat).Promotional pricing applies for 12 months from sign-up, and to charges after that date.

Read the full termsWhy scientists choose Claude "After collaborating with Claude Code for the past 3 months, turning one idea after another into research projects, I am totally hooked.I cannot imagine research anymore without it!

Claude's coding and reasoning capabilities are out of this world, and it's dramatically expanded my conception of what's possible.

"David Shih, Professor of Physics, Rutgers University“[Claude] was extraordinarily useful for my research -- both a larger, purely empirical project with a complicated codebase, and some more basic research projects and idea seeds where I needed to prototype something quickly.

It definitely changed the game for me, and I re-subscribed!"Hannah Lawrence, Computer Science Ph.D.

student, Massachusetts Institute of Technology"Claude has changed the way that I do science, from the way that I explore the literature and brainstorm for a new hypothesis, to our ability to analyze large-scale datasets and integrate diverse information sources at a scale that would be difficult to achieve otherwise.

It has accelerated my work, but has also allowed me to generate testable and verifiable ideas and to think about the science in fundamentally new ways.Claude passes the scientific “deletion test” - it has become an essential tool for our work.

”Iain Cheeseman, Herman and Margaret Sokol Professor of Biology; Core Member, Whitehead Institute; Associate Department Head“Claude has become a useful tool for our medicinal chemistry work.

It pulls structure-activity trends out of dense SAR tables and can read chemistry from images into SMILES and reaction steps.

We've even built a med chem agent using Claude that reasons over structures, computed and predicted properties, and experimental data to flag liabilities and surface which series are worth advancing.

”Akhila Kosaraju, CEO, Phare Bio“Claude has helped our lab identify promising research directions to hone in on by enabling rapid preliminary analysis and statistical power prediction for complex, highly specialized experimental designs.

”Silvia Domke, Assistant Professor, University of Zurich“We have taken data harmonization projects and standards development efforts that would have taken a decade to complete and realized them in months.

Claude has fundamentally changed how we interact with domain experts and the content we can deliver, such as for rare disease diagnostics and drug repurposing.

”Melissa Haendel, Director of Precision Health & Translational Informatics at University of North Carolina at Chapel Hill"Over the last six months, Claude has become a crucial collaborator for our lab, from designing computer processors, to programming languages, to operating systems.

As a team of computer systems builders, we can engage in high-risk, engineering-heavy prototyping which would be difficult to justify in the past.

"Jonathan Balkind, Assistant Professor of Computer Science, University of California, Santa Barbara“I use Opus as an intellectual partner to find holes in my ideas before I commit resources to experiments.At the bench, I generate protocols and troubleshoot logistics in real time.

With Claude Code, I automate data transformations and build analysis pipelines.As an MD-PhD student juggling course work, lab experiments, analysis, collaborations and scientific writing, having a tool this versatile has been transformative.

”Maia Madison, MD PhD Student, University of California, San Francisco"I’ve been able to take advantage of Claude’s broad knowledge to connect a research problem in theoretical particle physics to techniques developed for industrial systems engineering and number theory.

Claude Code has allowed me to identify and integrate well developed libraries into our niche problem setting and explore bespoke solutions that leverage advanced reasoning capabilities.It has dramatically changed the pace and ambition of this research project.

"Kyle Cranmer, Professor and Director of the Data Science Institute, University of Wisconsin, Madison"'I figured out how to do [insert X] with Claude' has become a frequent refrain in discussions with my team.

Claude is enabling my academic research group to acquire new skills and accomplish analysis tasks far more quickly.

"Abby Buchwalter, Assistant Professor, University of California, San Francisco PrevPrev0/5NextNextMore resourcesGet started with Claude team for scientistsGet started with Claude team for scientistsGet started with Claude team for scientistsLearnLearnLearnSee Claude Science in actionSee Claude Science in actionSee Claude Science in actionDemoDemoDemoLearn about the AI for Science grant programLearn about the AI for Science grant programLearn about the AI for Science grant programLearnLearnLearnHow Claude is accelerating protein design and analytical chemistryHow Claude is accelerating protein design and analytical chemistryHow Claude is accelerating protein design and analytical chemistryLearnLearnLearnVibe physics: The AI grad studentVibe physics: The AI grad studentVibe physics: The AI grad studentLearnLearnLearnMaking Claude a chemistMaking Claude a chemistMaking Claude a chemistLearnLearnLearnFAQWho is eligible to apply?

Principal investigators (PIs) or equivalent at accredited universities and nonprofit research institutes working in the natural sciences, mathematics, computer science, engineering, and related fields are eligible to apply.

If you are a for-profit company, contract research organization, or industry R&D team, please see our Team and Enterprise plans.What’s included in this discounted Claude Team plan?

The plan includes everything in Claude Team: shared projects, Claude Science, Code, Cowork, file creation, built-in connectors to research databases, single sign-on, and central billing and administration with a minimum purchase of 1 seat.If I don’t qualify, which plan is right for me?

If you're at a for-profit organization or don't meet the eligibility criteria, Claude Team and Claude Enterprise offer the same capabilities at standard pricing.Visit claude.com/pricing to compare plans, or contact our sales team for help choosing the right fit.

How does verification work, and how long does it take?You'll confirm your institutional affiliation and tell us briefly about your research.Most applications are reviewed within 5-7 business days.You'll get an email when you're approved and can add seats right away.

What does it cost and how many seats can I have?Standard seats are free.Premium seats are $15 per month.This is promotional pricing for verified research groups, available for 12 months from when you sign up (or from Thursday, August 27, 2026) and offered to a limited number of groups.

After 12 months, seats renew at the then-current price for this plan.We will notify admins before that happens.What is AI for Science, and how does it relate to this plan?AI for Science is Anthropic's grant program for ambitious research projects that need more credits than a seat provides.

Groups on this plan can apply for extra usage or API credits to support months of sustained work.It's application-based and reviewed on merit.What if I stop using it?If a group has no activity for 90 days, we may return it to standard Team pricing so the promotional seats go to active researchers.

We'll email the admin before that happens.Does Claude train on my data?No.Claude does not train on your conversations, uploaded files, or research data.Your unpublished findings, grant drafts, and proprietary datasets stay private.Admins can configure data retention policies for their team.

PrevPrevNextNext

Related

相關文章

MarkTechPost AI生成式AI

認識 Voxel51 3.0:為實體 AI 打造的統一多模態資料平臺

Voxel51 已發布 Voxel51 3.0,這是其資料平臺的一次重大更新。Voxel51 3.0 是一個端對端的多模態資料平臺,將搜尋、篩選、標註和評估整合於單一系統中。該平臺鎖定針對機器人、車輛和工廠感測器資料進行模型訓練的團隊。此次發布新增了 Voxel51 Agent、資料溯源層及免費的自助式存取。重點摘要:規模:Voxel51 Agent 內建超過 45 個 MCP 工具、80 多個運算子及 18 個開源技能。執行環境:可於 app.voxel51.com 使用託管網頁應用程式,或於雲端、混合、地端及隔離環境中自行部署。未揭露本機 GPU 需求。效能:未發布產品基準測試。Voxel51 引用其自身調查數據作為依據。最佳亮點:免費方案包含 Voxel51 Agent、每月 30,000 次推論請求及 30 個運算小時。

21 小時前
何夕2077生成式AI

下一概念建模把文本生成提速最高31倍

斯坦福等聯合提出全新文本生成範式 NCM,生成速度狂飆 31 倍AIbase基地發佈於AI新聞資訊 · 1 分鐘閱讀 · Oct 16, 202633人工智能領域的文本生成技術迎來顛覆性創新。斯坦福大學與 Together AI 的研究人員近日聯手推出了一種截然不同於傳統自迴歸和擴散模型的全新文本生成範式——Next-Concept Modeling(簡稱 NCM,即“下一概念建模”)。該方法巧妙借鑑了圖像生成領域中“下一尺度預測”的思路,有望為未來的文本生成技術開闢全新的發展路線。在技術實現路徑上,NCM 模擬了人類“先構思宏觀大綱、再動筆填充細節”的寫作邏輯。整個系統通過兩階段訓練來完成:首先訓練多尺度殘差 VQ-VAE,從而得到能夠逐層補充信息的多層離散文本編碼;隨後凍結該系統,訓練 Transformer 模塊去預測下一層的編碼。在生成文本時,模型會先並行生成概念層面的宏觀大綱,接著逐層並行填充具體細節,最終組合成完整的文本內容。在實際測試表現中,NCM 交出了令人矚目的成績單。實驗數據顯示,在保持相近甚至更優生成質量的前提下,NCM 相對傳統的擴散語言模型實現了10到31倍的生成速度飛躍,並且在訓練過程中所需消耗的 token 數量也大幅減少。此外,在數獨等複雜推理任務上,它的表現也明顯優於傳統的自迴歸和擴散基線模型,同時還支持通過多生成候選挑選機制來進一步優化最終輸出效果。儘管目前該研究所使用的模型參數規模依然相對有限,其在擴大模型規模後的表現、精確文本以及長篇內容的生成能力仍有待後續驗證,但這一創新思路無疑打破了長期以來的技術慣性,為人工智能高效生成文本提供了充滿想象力的全新解法。本文來自AIbase日報掃碼查看歡迎來到【AI日報】欄目!這裡是你每天探索人工智能世界的指南,每天我們為你呈現AI領域的熱點內容,聚焦開發者,助你洞悉技術趨勢、瞭解創新AI產品應用。

1 天前
雷峰網生成式AI

12.88萬起、華為電驅+L4同源智駕,Ray7能成為埃安年輕化的答卷嗎?

12.88萬起、華為電驅+L4同源智駕,Ray7能成為埃安年輕化的答卷嗎? 本文作者: 倪萍 2026-10-17 22:42 導語:10月17日,埃安轎跑Ray7正式開啟預售,預售價12.88 萬起。新車車長4.9米,軸距2.9米,車頭部分搭配3.25米的超長雙環燈帶,是埃安10年煥新的家族 10月17日,埃安轎跑Ray7正式開啟預售,預售價12.88 萬起。新車車長4.9米,軸距2.9米,車頭部分搭配3.25米的超長雙環燈帶,是埃安10年煥新的家族式標識。 新車採用寧德時代電池以及廣汽自研的彈匣電池2.0技術,能夠在充分釋放續航能力的同時,監控電池健康狀態。提供45.58kWh和62.98kWh兩種規格,續航可達702公里,整車風阻係數 Cd 低至 0.22。Ray7搭載華為DriveONE多合一電驅,以及華為DriveONE芯片級驅制融合架構,百公里制動僅為32.8米。底盤方面,Ray7搭載高性能前後穩定杆和後五連桿懸架,過彎極限更高,濾振更乾淨。智能化上, Ray7搭載27顆車規級全視角傳感器,採用文遠知行L4級Robotaxi同源大模型。這幫助新車能夠複用 L4 積累的海量場景經驗,在城區掉頭、繞行等複雜場景中可順利通行。自2025年廣汽推進“番禺行動”以來,昊鉑埃安BU的研發、生產、銷售等環節效率不斷提升,產品迭代和新車投放節奏明顯加快。作為埃安全新Ray車系的首款新車,Ray7是埃安開啟年輕化序列的關鍵落子,也是廣汽切入20萬級市場的重要佈局。這一細分市場競爭並不輕鬆:小鵬P7、極氪007等車型已經提前卡位。Ray7想在年輕人心中佔位,不能只堆參數,更要把智能座艙與輔助駕駛、三電安全、補能效率、用車成本和價格權益上的差異,變成用戶可感、可信、可持續的體驗,這才是埃安年輕化的分水嶺。(雷峰網(公眾號:雷峰網)) 雷峰網原創文章,未經授權禁止轉載。詳情見轉載須知。

1 天前
智東西生成式AI

梁聖又掀桌了!DeepSeek新模型斬殺GPT-6,許願式編程時代降臨

作者 | 楊京麗 編輯 | 李水青 10月17日報道,剛剛,DeepSeek V4.1 Flash更新至V4.1-Flash-1017版本,官方網頁端、App與API已同步更新至新模型,模型權重也已在HuggingFace和ModelScope開放下載。DeepSeek還向公眾開放自研搜索引擎的Search(搜索)與Fetch(抓取)API,兩個接口均定價為5元/千次。 此次更新距離該模型發佈僅過去一個多月。9月10日,DeepSeek正式推出V4.1-Flash,在性能上超越了DeepSeek-V4-Pro。

1 天前