- September 10, 2026
- FOXITBLOG
Highlights
- Everything stays on-device: Chat, editing, summarizing, and translation all run locally. Nothing crosses the network.
- Ties into the same release as BYOM and Workspace: Local AI and Bring Your Own Model both control where AI processing happens for work inside a single document. Foxit Workspace handles the separate case of work spanning many documents.
- Not a stripped-down mode: The same chat, editing, summarizing, and translation available in the cloud, just running locally instead.
- Built for IT to verify: Nothing to take on faith. IT can confirm exactly where processing happens by checking outbound network calls.
- One-time setup, then fully offline: Download a matched model once, switch to On-Device mode, and work without a network connection from there.
- Included, not sold separately: Available through the AI subscription add-on on Foxit PDF Editor, PDF Editor+, and Foxit PDF Reader, rather than as its own product.
Foxit PDF Editor 2026.2 brings a handful of new AI capabilities into the editor itself: Local AI, Bring Your Own Model, Foxit Workspace, and a redesigned AI Assistant. Local AI is the piece built specifically for documents that can’t leave the building at all, a contract, a patient record, a case file, anything where sending content to the cloud isn’t an option in the first place.
What Foxit Local AI Is
Local AI runs supported, text-based tasks directly on your device through Microsoft Foundry Local, once a matched model is downloaded. Chat, editing, summarizing, and translation all work the same way they would in the cloud. This isn’t a limited or stripped-down version built for offline use. It’s the same functionality, just running somewhere your IT team can actually verify.
Why It’s Part of This Release
Bring Your Own Model connects Foxit to an external endpoint you choose and control. Local AI goes a step further and removes the external endpoint entirely, keeping everything on the device itself. Foxit Workspace, in the same release, solves a different problem: work that spans many documents at once rather than one file in isolation. Together, the three features cover different points on the same spectrum, how much control a team needs over where AI processing happens, and how many documents it needs to touch at once.
On-Device AI Chat and Edit PDF Agent
Conversational Q&A on an open document runs entirely on-device, so a question about a contract or a case file never has to leave the machine it’s opened on. The same is true for the Edit PDF Agent: natural-language PDF editing, including redaction, watermarking, reordering, and page numbering, processed entirely on-device from a plain-language instruction.
On-Device Summarization and Translation
Document summaries and full-document translation are generated entirely on-device for supported text tasks. A summary or a translation comes back the same way a cloud-based one would, just without the file ever leaving the machine it started on.
How It Works
Setup is a one-time step, then everything runs offline from there. First, download a matched model through Microsoft Foundry Local. Second, select On-Device under AI Settings, under Models. Third, work fully offline: chat, edit, summarize, and translate, with no separate onboarding required after that first model download.
Local AI vs. Bring Your Own Model
The two features solve overlapping but different privacy needs. Local AI keeps everything on your device, full stop, with no external endpoint involved at all. Bring Your Own Model connects to a self-hosted or third-party endpoint you choose over an OpenAI-compatible API. Foxit doesn’t receive your content either way, but with BYOM, privacy from that point on depends on the endpoint you connected, not on Foxit. Local AI runs on qualifying Windows and Mac desktop devices in Foxit PDF Editor and Foxit PDF Reader only, not on cloud, mobile, or inside Foxit Workspace. BYOM has no hardware qualification requirement, since the processing happens on your chosen endpoint rather than the local device.
Who It’s For
Local AI is built for regulated legal, financial, healthcare, and government teams, along with IT teams that need to verify files stayed on-device, and organizations with an existing model license through BYOM.
Local AI FAQ
What is Foxit Local AI?
Foxit Local AI runs supported chat, editing, summarization, and translation tasks entirely on your device, so a file’s content never reaches the cloud.
Can I trust that Local AI really doesn’t send anything to the cloud?
Yes. Once a device-matched model is downloaded through Microsoft Foundry Local, supported tasks run entirely on that device, and IT can confirm it by checking outbound network calls.
What devices support Local AI?
Qualifying Windows and Mac devices running Foxit PDF Editor or Foxit PDF Reader as desktop applications. It isn’t available on cloud, mobile, or inside Foxit Workspace.
What doesn’t work in Local AI mode?
Agent automation, MCP connections, Skills, image generation, and podcast creation stay cloud-only. Local AI covers chat, edit, summarize, and translate.
What is Bring Your Own Model, and how is it different from Local AI?
BYOM connects Foxit to a self-hosted or third-party-hosted model you choose over an OpenAI-compatible API. Local AI keeps everything on your device with no external endpoint at all; with BYOM, content goes to the endpoint you selected instead, and Foxit never receives it either way.
How much does Local AI cost?
It’s included in the AI subscription add-on available on Foxit PDF Editor, PDF Editor+, and Foxit PDF Reader, rather than sold as a separate product.
Foxit Local AI is available now in Foxit PDF Editor and Foxit PDF Reader, part of the same 2026.2 release bringing Bring Your Own Model and Foxit Workspace into the platform.
Try it in a document you already have open, or contact sales to learn more about qualifying hardware and rollout for your team.

