BastionGPT vs DAX Copilot and Dragon Copilot (Nuance Dragon): AI Medical Scribe Comparison for 2026

Price you can see, and a total cost that is a fraction of Dragon's

Microsoft publishes no subscription price for Dragon Copilot. Its official licensing guidance directs buyers to a Microsoft representative or a Cloud Solution Provider partner for a quote, and the Microsoft Marketplace listings for DAX Copilot and Dragon Medical One require pre-purchase coordination with DragonMarketplace before an organization can even buy. The concrete numbers that are public come from authorized resellers: as of July 2026, DictationOne lists Dragon Copilot Physician Practice licensing at $159.99 per user per month with a required 12-month term, a $350 one-time implementation fee, and a $175 annual account maintenance fee. Historic pricing for the same product line ran higher: reseller listings and industry reporting placed DAX Copilot between $369 and $830+ per provider per month before Microsoft's May 2026 price restructuring, and older reviews still cite those figures.

BastionGPT publishes every price. Professional starts at $20 per user per month. Professional Plus, the plan most comparable to a full Dragon Copilot license, costs $45 and includes the full AI assistant, multi-document analysis, image and spreadsheet reading, and manual model selection. There is no implementation fee, no maintenance fee, and no required contract. Prices have held steady for over two years, and annual billing includes a free month.

The math for a five-clinician practice, using the reseller's published figures: Dragon Copilot runs $9,599 per year in subscriptions ($159.99 x 5 x 12), plus $1,750 in first-year implementation fees and $875 per year in account maintenance, for a first-year total near $12,224. BastionGPT Professional Plus runs $2,700 ($45 x 5 x 12). That is a difference of more than $9,500 in year one, and roughly $7,770 every year after.

‍

No meters, no Azure billing plan, no session math

Dragon Copilot's lower-cost license, Physician Flex, covers front-end dictation only. Ambient recording and generative AI are metered: each Flex license includes 10 AI-Assisted Sessions per month, pooled across the organization, with unused sessions expiring at month end. Anything beyond the allotment bills per session through a linked Azure subscription, which the organization must set up with its own resource group and billing plan. Microsoft's licensing guidance notes that an AI-Assisted Session is consumed by ambient note generation, chat queries, referral letters, and other generative features. A clinician who sees 20 patients a day exhausts a month of included sessions before lunch on day one.

BastionGPT includes unlimited AI Scribe sessions and unlimited AI assistant use on every plan, including the $20 Professional plan. There is no session meter, no consumption unit, and no cloud billing plan to configure.

‍

A full AI assistant that reads your documents

Dragon Copilot's generative features are anchored to the encounter: it drafts referral letters, after-visit summaries, evidence summaries, and orders from the visit, and its chat can query shift transcripts and search clinical information. Uploading and analyzing outside documents is not a documented capability, and spreadsheet analysis does not appear anywhere in Microsoft's product documentation.

On Professional Plus and Ultra plans, BastionGPT reads 10+ documents at once, up to 1,000+ pages of combined context. Supported formats include PDF, Word, Excel, CSV, PowerPoint, HTML, plain text, and JPG or PNG images, including scanned PDFs, charts, and graphs. A clinician can upload a full patient history, a specialist evaluation, and records from multiple providers, then query, summarize, and cross-reference all of it in one conversation. The assistant also drafts insurance appeal letters, prior authorizations, treatment plans, and patient education materials; reviews documentation for incorrect names, pronouns, vague language, and suspected under- or over-coding; and analyzes lab summaries and spreadsheets. BastionGPT achieved 99.3% sentence-level accuracy on clinical documentation tasks from a peer-reviewed study by Baylor College of Medicine and Texas Children's Hospital researchers.

‍

A scribe that knows who is speaking

BastionGPT recognizes up to 10 speakers per session, live or from uploaded audio, with sessions up to 240 minutes. Speaker diarization separates the transcript by voice, and BastionGPT's AI automatically labels each speaker's name and role (clinician, patient, parent, interpreter) in the transcript. It captures telehealth visits through Zoom, Teams, and Google Meet with no visible bot in the meeting. That matters for family sessions, group therapy, care conferences, and any visit with more than two voices in the room.

Dragon Copilot supports multi-party conversation capture but publishes no speaker count, and Microsoft's support documentation lists a 120-minute maximum recording time for the DAX Copilot mobile app, half of BastionGPT's 240-minute ceiling. Microsoft's own release notes state that when multilingual recording is enabled, speaker separation in the US is available only for English and Spanish, and in Canada the multilingual output arrives as a single paragraph with no speaker identification at all.

‍

Notes in your format, your specialty, and your voice

BastionGPT produces SOAP, Detailed SOAP, DAP, BIRP, H&P, intake assessments, progress notes, and treatment plans as first-class outputs, and a Custom Note option accepts any free-form prompt per recording. A single recording generates six output tabs at once: transcription, summary, SOAP, Detailed SOAP, DAP, and a custom note. Upload sample notes and BastionGPT matches your writing style, so the draft reads like you wrote it.

Dragon Copilot's customization is real but differently shaped. The clinician's primary specialty determines which AI model generates the note, and customizable templates are section-based: Microsoft documents template customization for Allergies, Physical Exam, and Assessment and Plan in the DAX Copilot for Dragon Medical One workflow, with the Epic workflow adding History of Present Illness, Results, and Procedure. Template configuration happens at the organization level, so an individual clinician adjusts within the sections IT has set up.

‍

A strong fit for mental health and multidisciplinary teams

BastionGPT publishes dedicated therapy and behavioral health workflows: DAP, BIRP, and custom therapy notes, intake assessments, and treatment plans, with 10-speaker recognition that handles group therapy and family sessions cleanly. It works alongside the systems mental health practices actually run, including SimplePractice, AdvancedMD, and Kareo, as well as hospital EHRs.

Microsoft's public DAX and Dragon materials center on physician encounter documentation and clinical summaries. For a practice with therapists, psychiatrists, counselors, and case managers on one roster, BastionGPT is one platform that covers the entire organization.

‍

AI models you can see and choose

BastionGPT names its engines: licensed and HIPAA-compliant versions of GPT-5.5, Claude, and Gemini 3 Pro. Every plan selects the best model for each task automatically, and Professional Plus adds manual model selection plus early access to new models as they ship. Model quality drives note quality, and naming the models lets you verify exactly what is reading your records.

Microsoft describes Dragon Copilot as running on "fine-tuned generative AI with healthcare-adapted safeguards." It does not name the model versions in use, and clinicians cannot select a model.

‍

Your data is never training data

This difference is worth reading twice. Microsoft's Dragon Copilot FAQ states that audio, transcript, and flowsheet data is retained for up to 90 days, and that data is deidentified or pseudonymized for AI model improvement. Your patient encounters, in deidentified form, help train Microsoft's models.

BastionGPT never uses your prompts, documents, or audio to train any AI model, deidentified or otherwise. Chats and transcripts are wiped after 30 days by default, and you can retain them longer or delete them sooner. Customer data is segregated by tenant, encrypted with AES-256 at rest and TLS in transit, runs on HITRUST CSF certified and SOC 2 Type II attested Microsoft Azure infrastructure, and is independently penetration tested at least annually. BastionGPT enters a signed Business Associate Agreement with every customer on every plan, including the free trial; the BAA is built into the standard terms of service, and a countersigned copy is available on request. BastionGPT supports HIPAA, 42 CFR Part 2, FERPA, PIPEDA, PHIPA, and Australian APP requirements. In three years serving more than 10,000 healthcare organizations, BastionGPT has had zero breaches of protected health information.

‍

Buy it today instead of procuring it this quarter

Dragon Copilot is purchased through a Microsoft Customer Agreement, a CSP partner, or an Enterprise Agreement, and the Marketplace listings require pre-purchase eligibility verification through DragonMarketplace. Deployment requires Microsoft Entra ID for authentication, provisioning through the Dragon admin center, and, for Flex licensing, a linked Azure subscription with a billing plan. The legacy DAX Copilot for Dragon Medical One workflow adds its own dependencies: recording runs through PowerMic Mobile on iOS, and note transfer to the EHR happens through Dragon Medical One on desktop. Independent reviews of the DAX product line describe enterprise implementation timelines of 3 to 6 months, and the reseller channel bills a per-user implementation fee before the first note is generated.

BastionGPT deploys in minutes. A solo practitioner or a 50-clinician group signs up online, signs the BAA electronically, and starts documenting the same day. BastionGPT works alongside 50+ EMR/EHR systems, including Epic, Oracle Health, Athenahealth, eClinicalWorks, SimplePractice, and AdvancedMD, via simple copy-paste, with no integration project and no IT department required. A Saved Prompts library with folder organization and team sharing is included on every plan, so a practice can standardize documentation across all providers from day one.

‍

What the independent evidence shows

BastionGPT holds a 4.8 out of 5 average across the three major verified B2B review directories: Capterra, Software Advice, and GetApp. Reviewers most often cite time saved on documentation and the freedom to use AI on real cases because data protection is built in.

Dragon Copilot is sold enterprise-only, so the most rigorous public evidence comes from independent research rather than review directories. A randomized clinical trial at UCLA Health, published in NEJM AI in November 2025 and funded by the university rather than any vendor, assigned 238 physicians across 14 specialties to DAX Copilot, a competing scribe, or usual care across roughly 72,000 patient encounters. The competing scribe reduced time-in-note by a statistically significant 9.5% versus control. DAX Copilot's reduction was smaller and did not reach statistical significance. Both tools produced modest improvements in burnout measures, on the order of 7%. Physician communities echo the mixed picture: recurring complaints include verbose assessment-and-plan sections that need trimming and pricing widely described as out of reach for anyone outside a large health system.

Figures shift over time, so this page cites the source and date for each number. The links in the sources section lead to the live figures.

‍