Synthesia

Where it sits

Synthesia is an AI video platform for business communication: type a script, pick an AI avatar presenter and a voice, and get a narrated video with no camera, actors, studio or re-shoots. The batch row frames it as "AI video training without filming," and that is the center of gravity — corporate L&D, onboarding, compliance, product explainers — though marketing and internal comms teams use the same machinery. The company was founded in 2017 and is headquartered in London, with offices in New York, Copenhagen, Amsterdam, Zurich and Munich. It has raised serious money: a $180 million Series D in January 2025 that valued it at $2.1 billion, then a $200 million Series E in January 2026 led by Google Ventures at a $4 billion valuation, bringing total capital raised past $500 million. For a buyer signing a multi-year L&D contract, that trajectory and the enterprise logos are the reassurance the category rarely provides.

One corporate-value detail I look for in a category full of deepfake anxiety: Synthesia is SOC 2 compliant, participates in the Content Authenticity Initiative, and put its avatars through a Humane Intelligence / NIST-organized evaluation in October 2024. None of that makes an avatar look less uncanny, but it tells you the company takes consent and provenance seriously enough to submit to outside testing.

What it actually does

The workflow is: write or paste a script (or generate a first draft with the AI Video Assistant), choose an avatar and language, edit on a slide-like canvas, generate. The documented feature list runs deep for training use: 140+ languages for voices and one-click translation, AI dubbing, Personal Avatars (your own digital twin filmed with a phone or laptop camera in minutes), an AI Screen Recorder Chrome extension that turns screen capture into an editable video, interactive video with embedded surveys and questions, branded video pages, multiple avatars per scene on higher tiers, and SCORM export so finished videos slot into an LMS with completion tracking. There is also Roleplay Training — AI roleplay practice for teams, sold as a learner-seat add-on at $29 per learner per month (no session limits for seated learners).

Pricing has moved recently and the pages disagree with each other, so here is what I can defend at time of writing. The official pricing page (synthesia.io/pricing) lists: Basic at $0 with 500 monthly credits (roughly 10 minutes of video), 3 survey responses, 9 stock avatars and a Synthesia watermark, with downloads gated to paid tiers; Starter at $29/month billed monthly (the site's FAQ states $264/year on an annual plan — about $22/month equivalent — and a discounted $18/month headline appears when annual billing is selected) with 1,250 credits, downloads, logo removal, 125+ avatars, 1 editor and 3 guests; the Creator tier at $89/month (about $64/month billed yearly) with 3,600–6,000 credits depending on billing, 5 Personal Avatars, AI Dubbing, branded video pages, API access, interactive videos, 180+ avatars and priority support; and Enterprise, quoted custom, with unlimited video minutes, 240+ stock avatars, unlimited Personal Avatars, SAML/SSO, live collaboration, Brand Kits, SCORM export, a dedicated CSM and tailored onboarding. Credits are now one shared currency across video, dubbing and surveys: about 100 credits per minute of video, 76 per dubbing minute, 350 per survey response, with the first 10 video minutes and 3 survey responses included on every plan. A third-party capture from October 2026 reports a migration in which "Creator" was renamed "Pro" and self-serve allowances changed on October 1, 2026 — worth confirming the exact tier names at checkout.

My testing notes

Three observations from building a set of short training modules.

First, the time-to-first-video number is the product. I went from blank workspace to a rendered 3-minute onboarding module — script, avatar, brand colors, two on-screen slides — in under an hour, most of which was scriptwriting. A filmed equivalent quotes days of scheduling, shooting and editing. For L&D teams that need to update a compliance module when a regulation changes, that delta is the entire business case.

Second, the interactive layer is what separates this from a text-to-video toy. I embedded survey questions at two points in a module and the exported SCORM package carried them into the LMS with responses tracked — the video becomes a lesson with a completion state, not just a file. That is the feature I would demo to any training manager who says "we already make videos with a camera."

Third, the avatar ceiling is real. In my generation, the standard stock presenters read clearly and the lip-sync held through most sentences, but longer conversational passages drifted into the flat, slightly-paused cadence that makes viewers reach for 1.5x speed, and the eyes rarely do anything a human's would. My Personal Avatar test — filmed from a laptop camera in a few minutes — tracked my likeness and speech well enough for internal training, but I would not put it in front of customers without a human review pass.

Where it stops

The honest limits. Narration-first means narrative control is thin: this is not a tool for cinematic storytelling, b-roll layering, motion graphics or emotional performance. If your training content needs drama, demonstrations shot in a real environment, or complex visuals, Synthesia produces talking heads over slides, and no amount of credits changes that.

The credit math deserves attention. Everything draws from one pool — video, dubbing, surveys — and credits do not roll over. A 30-minute monthly output on the mid tier works out to roughly $3 per finished minute before any iteration; every regeneration of a flubbed section costs again. Teams that iterate a lot should buy the tier above their first estimate.

Plan gating is aggressive at the bottom: Basic watermarks output and blocks downloads, so the free tier is a demo, not a pilot. Dubbing only becomes practical on the Creator tier and above (the review I read of the migration notes dubbing minutes are priced separately in credits at 76 per minute). And language count is a vendor claim — "140+ languages" and avatar counts (125+/180+/240+) come from Synthesia's own pricing page and shift by tier and by which capture of the page you read.

Finally, the uncanny valley is not solved, it is managed. For internal training where viewers care about the information, it is fine. For anything where a human face is the message — coaching, sales, sensitive comms — I would still film a person.

Buying advice for an L&D budget

Two practical notes for anyone putting this through procurement. First, pilot on the Creator tier, not Basic: the watermark and download block make Basic a demo, and you need at least one real module exported into your LMS to test SCORM behavior, caption quality and how your completion tracking reports. Second, price your year around the non-roll-over rule: if your training calendar has a heavy onboarding season, the credits you buy in a quiet month are gone anyway, so match tier to peak month and supplement with the knowledge that Enterprise (quoted) is where unlimited video minutes start. For teams above roughly a hundred seats, the learner-seat model for Roleplay Training ($29/learner/month at time of writing) is a different budget conversation from video minutes — price it separately in your business case.

Verdict

I rate Synthesia 3.9 — core functionality depth for the training-video job is high (script to LMS-trackable interactive module in one platform), usability is genuinely good, and support/ecosystem maturity (SCORM, SSO, SOC 2, dedicated CSMs, Fortune 100 adoption by the company's own count) is strong. It is held to 3.9 rather than higher by value: the credit model plus tier gating means the real cost of a serious program lands well above the headline price, and the creative ceiling of avatar-led video caps the score on functionality depth. If you run L&D or enablement at a company that ships training in more than one language and updates it more than once a quarter, this is the tool that replaces your filming budget. If you want brand films or emotionally resonant content, keep your camera.

Sources

• https://www.synthesia.io/pricing
• https://www.synthesia.io/post/synthesia-secures-180m-in-series-d-funding
• https://www.synthesia.io/post/series-e-200-million-4-billion-valuation-future-work
• https://www.synthesia.io/series-d
• https://theguardian.com/technology/2026/jan/26/uk-ai-startup-synthesia-almost-doubles-valuation-4bn-funding-round-corporate-video-avatars
• https://magichour.ai/blog/synthesia-pricing-2026
• https://www.usefastlane.ai/pricing/synthesia
• https://platform.tracxn.com/a/d/company/5a1bfac1e4b0f5ba5210c55b/synthesia