Hedra
Character video AI: a portrait plus audio becomes a directed talking or acting performance
Hedra turns a character image (photo, illustration or 3D render) and an audio track or text prompt into video, with lip-sync, facial expression and body motion. Its Character-3 and Avatar models are the core, Omnia is its own unified audio-visual model, and the Hedra Agent plus shared Spaces handle multi-step creative work. It also hosts third-party models such as Kling, Veo and Seedance behind one account and API.
- 4 creators using
- 5 news items tracked
- 2024 first launched
- Freemium pricing
- Easy learning curve
What Hedra can do
- Infinite canvas Yes
- AI agent Yes
- Node / flow builder Not confirmed
- Public API Yes
- MCP server Yes
18 of 26 capabilities confirmed
| Capability | Supported | Details |
|---|---|---|
| Text to image | Yes | Hosted image models (GPT Image 2, Nano Banana Pro/2, Seedream 5.0 Lite) |
| Image editing | Partial | Restyling and reference-based edits through Skills and hosted image models |
| Text to video | Yes | Hosted third-party video models; Omnia and Character-3 are image- or audio-driven |
| Image to video | Yes | Character-3, Hedra Avatar and Omnia from a reference image |
| Video to video | Partial | Agent Video Edit mode (Aug 2026); scope not fully documented |
| Native audio in video | Yes | Omnia generates audio-visual output together; hosted Veo and Kling models also generate audio |
| Lip-sync | Yes | Character-3 and Hedra Avatar are built for audio-driven lip-sync |
| Voice / TTS | Not confirmed | |
| Music & SFX | Not confirmed | |
| Upscaling | Not confirmed | |
| Character consistency | Yes | Character identity comes from the reference portrait in Character-3 and Avatar |
| Camera control | Yes | Camera control in Hedra Avatar and Omnia (vendor claim) |
| Custom training | Not confirmed | |
| Third-party models | Yes | Hosts Kling, Veo, Seedance, Sora, Grok Imagine, MiniMax and others; 110+ models via the API |
| Capability | Supported | Details |
|---|---|---|
| Infinite canvas | Yes | Shared Studio Space canvas with labelled collaborator cursors (Aug 2026) |
| Node / flow builder | Not confirmed | |
| AI agent | Yes | Hedra Agent (general creative agent since May 2026) |
| Chat assistant | Yes | Chat-driven Hedra Agent with reusable Skills |
| Team collaboration | Yes | Realtime multiplayer in Agent Spaces and shared canvas (Aug 2026) |
| Capability | Supported | Details |
|---|---|---|
| Public API | Yes | Hedra Develop API (generally available Aug 2026), base URL api.hedra.com |
| MCP server | Yes | Hedra MCP server at mcp.hedra.com (OAuth sign-in) |
| Mobile app | Not confirmed | |
| Desktop / local | No | |
| Open source | No |
| Capability | Supported | Details |
|---|---|---|
| Free tier | Yes | Free plan without a credit card (per the pricing page; allowance not verified) |
| Commercial use | Yes | Commercial use on paid plans, per the pricing page |
Yes Partial No Not confirmed by the vendor
What makes Hedra different
- 01
Built for character performance first: one portrait plus audio gives lip-sync, expression and gesture, and it works on illustrated characters as well as photos.
- 02
Character-3 clips run up to 10 minutes per generation, far longer than most single-shot video models, and are priced per second at 2.5c (540p), 5c (720p) or 6.25c (1080p).
- 03
Hosts other companies' models (Kling O3 and O1, Veo, Seedance 2.0, Sora 2 Pro, Grok Imagine, MiniMax) behind one account, so a filmmaker can switch models without a second subscription.
- 04
One API key and one wallet cover the whole catalog through the Develop API, and the MCP server lets Claude or Codex call Hedra directly.
- 05
Agent Skills and shared Spaces turn repeated briefs (storyboards, UGC, product showcases, character sheets) into reusable steps for a team.
Latest from Hedra
- 3w ago Developer API Opens Character Video Generation to External Apps Hedra launched a developer API, SDKs, and Model Context Protocol support to let creators and developers build AI character video generation directly into their own applications.
- Sep 25 Hedra releases API pricing breakdown for Seedance, Kling, Veo and MiniMax Hedra details how much 10 seconds of AI‑generated video costs on Seedance, Kling, Veo and MiniMax APIs.
- Aug 31 MCP Server and CLI Enable Automated Character Video Generation in AI Agents Hedra launched an MCP server and CLI, enabling developers to trigger character video generation directly from AI agents and IDEs like Cursor.
- Aug 23 Hedra Rolls Out Verification Framework for AI Video Generation Hedra introduces an AI verification framework to validate synthetic video generation and establish clearer digital content provenance.
- Aug 19 Unomundi Adopts Hedra AI Character Video Generation for Global Children’s Content Unomundi integrates Hedra’s AI video generation engine to create localized, character-driven educational content for children globally.
About Hedra
Hedra started as a character-animation tool: give it one portrait and an audio file and it produces a talking or performing clip. Character-3 is the model most creators use for that, with 540p, 720p and 1080p output, up to 10-minute clips and 1:1 through 21:9 aspect ratios. Hedra Avatar, released in June 2026, adds better lip-sync and camera control on top of Character-3. Omnia, Hedra's own omnimodal model, takes a reference image, an audio track and a text prompt and generates audio-visual shots with control over camera, expression and background.
The 2026 changes are mostly about the platform. Hedra Agent was rebuilt in May 2026 as a general creative agent that researches, plans and generates video, image and audio in one workspace. Skills (saved, reusable playbooks such as storyboards, UGC and campaign briefs) and realtime multiplayer in Agent Spaces arrived in August 2026. The Hedra Develop API became generally available in August and was documented on October 6, 2026, with one API key, a prepaid USD wallet, an MCP server at mcp.hedra.com, and more than 110 models in one catalog.
For a film or video workflow, Hedra fits dialogue-driven character shots: explainers, UGC-style ads, music videos with characters, animated presenters, and talking-head inserts where a full actor shoot is not practical. It is weaker as a general cinematic generator. Long narrative sequences, multi-character blocking and precise camera choreography still need a tool like Kling, Veo or Seedance, which Hedra also hosts.
What's new in Hedra
-
Hedra Developer API and MCP server documented
One API for 110+ models across video, image, audio and text, billed from a prepaid USD wallet, with an MCP server at mcp.hedra.com that signs in with OAuth.
Source: hedra.com ↗ -
August platform update: Agent multiplayer, Skills and Video Edit
Realtime multiplayer in Agent Spaces, a Personalize area for reusable Skills, a Video Edit mode in Agent, labelled collaborator cursors on the canvas, and USD billing with auto-recharge for API credits.
Source: hedra.featurebase.app ↗ -
Gemini Omni Flash Video and expanded Kling/Veo reference inputs
Added to the hosted model catalog in the same August update; Kling and Veo gained reference-media support.
Source: hedra.featurebase.app ↗ -
Hedra Avatar launches after Character-3
Takes one portrait and an audio track and produces a directed on-camera performance with better lip-sync and camera control.
Source: hedra.com ↗ -
Hedra Agent upgraded to a general creative agent
The agent now researches, plans and makes video, image and audio, and delivers finished output in one workspace instead of only prompt-level help.
Source: hedra.com ↗
Models available in Hedra
-
Veo 3.1 Latest $$$ Google Hosted; reference-media inputs expanded Aug 2026; Studio credit rate listed on Hedra plans page
-
Sora 2 Pro Latest — OpenAI Hosted; price corrected Aug 2026
-
Seedance 2.0 $$$ ByteDance Hosted; multimodal inputs, native sound
-
Kling O3 $$$ Kuaishou Listed as available on Hedra in mid-2026 Also: Kling O1
-
MiniMax H3 $ MiniMax Hosted via API; example price $0.36 for 6 s at 768p
-
Character-3 — Hedra Portrait plus audio to talking video; 540p/720p/1080p; up to 10 min per clip; 2.5c to 6.25c per second
-
-
-
-
-
Grok Imagine — xAI Hosted; price corrected Aug 2026
$ · $$ · $$$ = relative cost against other models of the same type, on fal.ai list prices. Cost inside Hedra depends on its own plans and credits.
Key Features
- Character-3 talking and performing video from one portrait and an audio track, up to 10 minutes per clip
- Hedra Avatar with lip-sync and camera control (June 2026)
- Omnia omnimodal model: reference image, audio and text in one generation
- Hedra Agent that plans and generates video, image and audio in one workspace
- Reusable Skills (storyboards, UGC, product showcases, campaign briefs, character sheets)
- Shared Studio Spaces canvas with realtime multiplayer
- Hosted third-party models (Kling, Veo, Seedance, Sora 2 Pro, Grok Imagine, MiniMax) under one account
- Develop API with 110+ models, one API key, prepaid wallet and auto-recharge
- MCP server at mcp.hedra.com for Claude Code and other MCP clients
- Output at 540p, 720p and 1080p in 1:1, 4:3, 3:4, 16:9, 9:16, 9:21 and 21:9
When to reach for it — and when to skip
Reach for it when…
- Strong lip-sync and expression from a single portrait, including illustrated characters
- Clips up to 10 minutes per generation, longer than most single-shot video models
- Per-second pricing on Character-3 makes short tests cheap to run
- Hosts several frontier video models, so one account covers model comparisons
- API and MCP access are now documented and generally available
Skip it when…
- Output tops out at 1080p on Character-3 and Omnia, below 4K-capable rivals
- Credits expire monthly and do not roll over, according to third-party reviews
- Billing and support complaints recur in third-party reviews; one cited a low Trustpilot score that was not verified
- Audio quality drives results; noisy or reverberant audio produces weaker lip-sync
- Pricing is inconsistent across sources and the plan structure has changed in 2026
Best For
✓ Ideal for
- Talking-character explainers and spokesperson videos from a single portrait
- UGC-style ads and product walkthroughs with an illustrated or AI-generated presenter
- Music videos and animated shorts where a character must sing or talk on camera
- Pre-visualizing dialogue scenes before booking a live actor
- Teams that want Kling, Veo or Seedance output alongside lip-sync from one account and API
✗ Not built for
- Feature-length or multi-character narrative scenes that need precise blocking
- Projects that must deliver 4K masters
- Offline or self-hosted generation
Working Tips from Filmmakers Using Hedra
- 01 Use clean, dry audio. Background noise and reverb hurt lip-sync more than any setting, so record or clean the voice before uploading.
- 02 Check the per-second credit cost in the generation control before a long render. Character-3 at 1080p costs 6.25c per second, so a 60-second clip is about $3.75.
- 03 Test concepts on Character-3 or Omnia first and only move to hosted Kling or Veo once the shot is locked, since credits drain quickly on the largest models.
- 04 Use the API estimate endpoint (POST /v3/models/{model}/estimate) to price a batch before running it.
Hedra pricing in 2026
| Plan | Price | What you get |
|---|---|---|
| Free | $0 monthly |
|
| Basic | $20 monthly |
|
| Pro | $50 monthly |
|
| Teams | $75 monthly |
|
| Ultra | $100 monthly |
|
| Enterprise | Custom custom |
|
Prices checked against Hedra's pricing page on 9 Oct 2026. Vendors change plans often, so confirm before you buy.
Hedra vs HeyGen vs Sync.so vs D-ID
| Hedra | HeyGen | Sync.so | D-ID | |
|---|---|---|---|---|
| Price | From $20/mo (Basic, 2,000 credits); Pro $50/mo; Ultra $100/mo; Teams $75/mo; Enterprise custom | Free (3 videos/mo, 1-min cap) · Creator $29/mo · Pro from $49/mo · Business $149/mo + $20/seat | Free (3 gens/month, 20s max); Hobbyist $5/mo; Creator $19/mo; per-second usage on top (sync-3 $0.107 to $0.133/sec by plan) | Trial with watermark, then from $5.90/mo (Lite) |
| Infinite canvas | Yes | Partial | No | No |
| AI agent | Yes | Yes | No | Partial |
| Chat assistant | Yes | Partial | Partial | Partial |
| Team collaboration | Yes | Yes | Partial | Not confirmed |
| Text to image | Yes | Yes | No | Partial |
| Image editing | Partial | Partial | No | Not confirmed |
| Text to video | Yes | Yes | No | No |
| Image to video | Yes | Yes | Yes | Yes |
| Video to video | Partial | Partial | Yes | Partial |
| Native audio in video | Yes | Yes | No | Partial |
| Lip-sync | Yes | Yes | Yes | Yes |
| Voice / TTS | Not confirmed | Yes | Yes | Yes |
| Music & SFX | Not confirmed | Yes | No | No |
| Character consistency | Yes | Yes | Partial | Partial |
| Camera control | Yes | Partial | No | No |
| Custom training | Not confirmed | Partial | No | Partial |
| Third-party models | Yes | Yes | Partial | No |
| Public API | Yes | Yes | Yes | Yes |
| MCP server | Yes | Yes | Yes | Not confirmed |
| Mobile app | Not confirmed | Yes | Not confirmed | Partial |
| Desktop / local | No | Partial | No | Partial |
| Open source | No | Partial | No | No |
| Free tier | Yes | Yes | Yes | Partial |
| Commercial use | Yes | Partial | Not confirmed | Not confirmed |
Each column comes from that tool's own researched page. "Not confirmed" means the vendor doesn't say.
Open the full Hedra vs HeyGen vs Sync.so vs D-ID comparison →Use Cases
Tags
Sources
Facts on this page were checked against Hedra's own pricing, product and changelog pages on , then cross-checked with independent coverage where vendor pages were silent. Anything we couldn't confirm is marked "Not confirmed" rather than guessed.
- Hedra pricing page hedra.com
- Hedra home page hedra.com
- The Hedra Developer API: models, SDKs, MCP and the console (Oct 6, 2026) hedra.com
- Hedra changelog: August 2nd platform update (Aug 3, 2026) hedra.featurebase.app
- Hedra Character-3 model page hedra.com
- AI models that defined early 2026 (Hedra blog, June 15, 2026) hedra.com
- Hedra Creative Studio and Skills hedra.com
- HeyGen pricing (comparison) heygen.com
- Sync pricing (comparison) sync.so
Spotted something out of date? Tell us and we'll re-check it.
Discussion
No comments yet — be the first.