The Reducto Roundup: r-1 mini is here 🚀

No dark-mode styles — this email renders the same in a dark client.

r-1 mini at $1 per 1,000 pages, our new parser evaluation framework PAIR, Reducto Office, and the latest in AI.


https://reducto.ai/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=header_logo

Happy Monday!
Last week, we launched r-1 mini
[https://reducto.ai/blog/reducto-r-1-mini-parse-model?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=r1_mini_intro],
bringing high-quality document parsing to $1 per 1,000 pages. This edition introduces PAIR
[https://www.linkedin.com/posts/aditabraham_pair-open-source-parsing-evaluation-framework-ugcPost-7525609083500908544-qy3M/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=pair_intro],
our open framework for evaluating document parsers on your own data, alongside updates to Reducto Office and Extract and a roundup
of the latest in AI.

🚀 Introducing r-1 mini

https://reducto.ai/blog/reducto-r-1-mini-parse-model?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=r1_mini_announcement
Read the announcement →
[https://reducto.ai/blog/reducto-r-1-mini-parse-model?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=r1_mini_announcement]
https://www.linkedin.com/feed/update/urn:li:activity:7524154614959689728/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=r1_mini_video
Watch the r-1 mini launch video →
[https://www.linkedin.com/feed/update/urn:li:activity:7524154614959689728/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=r1_mini_video]

Premium parsing at $1 per 1,000 pages.
We’ve focused on helping teams work with their hardest documents. With r-1 mini, we’re making it more affordable to work with all
of them.
In our evaluations, r-1 mini outperformed tested alternatives costing up to 20× as much. Adit shares the thinking behind it in our
launch video
[https://www.linkedin.com/feed/update/urn:li:activity:7524154614959689728/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=r1_mini_video].
Bring a few files and try it with free credits. Personal email addresses are welcome, too.

Try r-1 mini in Studio →
[https://studio.reducto.ai/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=try_studio]

r-1 is now generally available! For your most demanding documents, our flagship model offers our highest accuracy, lower latency
than r-1 mini, and word-level citations. It’s a good fit for complex charts, degraded scans, and mixed-language files.

Moving from another provider? Eligible teams can get up to $5,000 in migration credits
[https://reducto.ai/migrate?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=migration_credits].

📰 This week in AI
 * OpenAI reportedly raises $11B at a $1.25T valuation.
   [https://bloomberg.com/news/articles/2026-11-06/openai-completes-11-billion-portion-of-funding-at-1-25-trillion-value]
 * Anthropic explores how agents build lasting memories. [https://www.anthropic.com/research/components-of-agent-memory] Its new
   research examines “dreaming”: agents revisit past interactions between tasks to correct mistakes and organize what they’ve
   learned.
 * OpenRouter launches Auto and Auto Code routers. [https://openrouter.ai/blog/announcements/auto-and-auto-code-routers/] The new
   routers pick models based on the task, cost, and availability. OpenRouter reports up to 4× better benchmark performance and 3×
   faster p90 routing than its previous router.
 * NVIDIA introduces GR00T 2 for robot learning. [https://blogs.nvidia.com/blog/robotics-simulation-advancements-corl/] The new
   model learns from human and robot demonstrations, tackling tasks such as connecting cables and assembling small components. It
   arrives alongside Isaac Lab 3.0 and new open benchmarks.

🔎 Evaluate parsers on your own documents with PAIR
A parser can win a benchmark and still fail on your documents.
That’s why we’re introducing
[https://www.linkedin.com/posts/aditabraham_pair-open-source-parsing-evaluation-framework-ugcPost-7525609083500908544-qy3M/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=pair_social]Pairwise
AI Ranking (PAIR)
[https://www.linkedin.com/posts/aditabraham_pair-open-source-parsing-evaluation-framework-ugcPost-7525609083500908544-qy3M/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=pair_social],
our open framework for evaluating document parsers. It compares two parser outputs against the original page and judges which
better preserves its information, allowing ties.
Built while developing r-1 and r-1 mini, PAIR helps teams evaluate parsers on their own documents without creating
document-specific grading rules.

https://reducto.ai/blog/pairwise-ai-ranking-parse-evaluation?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=pair_blog

Read how PAIR works →
[https://reducto.ai/blog/pairwise-ai-ranking-parse-evaluation?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=pair_blog]

See how parsers handle your documents with ParseJudge
[https://parsejudge.com/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=parsejudge],
comparing outputs and page-level evaluations side by side.

📝 Try Reducto Office

https://reducto.ai/office?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=office_image_openaihttps://reducto.ai/office?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=office_image_examples

Install the plugin →
[https://www.reducto.ai/office-plugin?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=office_install]

Put your agent to work on the files you already use. With Reducto Office
[https://reducto.ai/office?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=office_page],
you can revise documents, update spreadsheets, and build presentations directly in ChatGPT and Codex.
It’s free and collaborative, with a live editor so you can review changes and refine them together.

A few more product updates
📄 What’s changing in Extract
Our new citations model links extracted answers to their source and is available to try now. It becomes the default on December 7,
alongside r-1 as Extract’s default parser. You can switch early or keep legacy parsing pinned while you test.
We’re also opening a beta for a lower-cost Deep Extract option. Sign up for the beta →
[https://docs.google.com/forms/d/e/1FAIpQLScfjOHdsdRZ0eW-yamZk0KrHSA5pO9xkYhpi8VhFoTLyKOilQ/viewform?usp=send_form&utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=deep_extract_beta]

⚙️ Built around your workflows
We kept seeing customers use Extract just to identify document types. It worked, but it was more machinery than the job needed.
So we built Classify: a faster, purpose-built endpoint that identifies documents at 0.5 credits per page. It’s already processed
nearly 1 million pages.
For work that can wait, Batch gives Parse and Extract jobs a 20% credit discount. Our engineers Kirthi, Alvin, and Erica explain
how both grew out of patterns they saw in customer workflows.

https://x.com/raunakdoesdev/status/2112582263070499171?a=oDGgKR0t3Kg_Xy-4NUQ&s=20&utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=classify_batch_video

Watch the walkthrough →
[https://x.com/raunakdoesdev/status/2112582263070499171?a=oDGgKR0t3Kg_Xy-4NUQ&s=20&utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=classify_batch_video]

📅 Events
✈️ Take Flight With Reducto · NYC
Tuesday, November 17 · 5:30–9 p.m. EST
Next stop: The Squadron NYC! Join fellow enterprise leaders for an invite-only evening of F-35 flight simulations, dinner, drinks,
and a preview of what’s next at Reducto.
Fly three missions, then compete in the Top Aces finale for a TopGun pilot jacket. Seats are limited.

https://luma.com/reductoai-fiw4?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=flight_event

Request your seat →
[https://luma.com/reductoai-fiw4?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=flight_event]

Non-Slop Hackathon with Pangram and Nous Research
Sunday, November 22 · Brooklyn, NY
Join us in Brooklyn for a day of building. Every participant gets $1,000 in Reducto credits, with a $1,000 cash prize from us up
for grabs.
Sign up for the hackathon →
[https://luma.com/enrexhqq?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=non_slop_hackathon]

Until next time,
The Reducto team

https://reducto.ai/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=footer_logo

Reducto is the complete agentic document platform for leading AI teams needing performance at enterprise scale.
Reducto, Inc. © 2026

https://x.com/reductoai?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=social_xhttps://www.linkedin.com/company/reducto-ai/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=social_linkedinhttps://github.com/reductoai?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=social_githubhttps://huggingface.co/reducto?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=social_huggingface

You are receiving this email because you opted in to receive updates from Reducto.
Reducto, 77 Geary Street, San Francisco, CA 94108
[https://reducto.ai/?utm_source=reducto_roundup&utm_medium=email&utm_campaign=reducto_roundup_3p&utm_content=footer_address]
Unsubscribe [https://app.loops.so/unsubscribe/cmw9wszc11gd901tr4nr8evhd/54dd4599e6d17b033e0bdea81e4668cade715eb8623c2b73bf9ef8d1a9ab5a4308a336a20fb1029829d49fdbe7ee8ef344c0ab87b325c94e59d3381c]

Accession

nse/reducto/2026-11-09/6487

From

Vibha @ Reducto · vibha@hey.reducto.ai

Sent

9 November 2026 · 23:00 UTC