Skip to content

Product updates

FreeInd product changelog

What shipped, when. API breaking changes are announced here with a 90-day deprecation window.

Last updated November 19, 2026RSS feedAPI deprecation policy

November 2026

4 updates
  1. 2026-11-19

    • Attach up to ten saved free.character identities to a set and build a real shot list — establishing, group, two-shot, single and insert
    • The set can be a saved free.location plate (with its light and angle plates), an image from your library, a file you upload, or a pasted URL
    • Direct each shot: framing, camera angle, lens, expression, pose, the action beat and per-character blocking (placement, posture, gaze), plus outfit overrides that keep the face and hair locked
    • Every render sends the plate as the source image and each character’s identity as a reference; a cast larger than the model can hold is refused up front instead of silently dropping a face
    • A plate-vs-shot lens or framing conflict is reported before credits are spent, so a close-up never quietly crops the set you designed
    • Animate any rendered shot with camera-motion presets and per-second durations; models that support a last frame can interpolate from one shot’s still into another
    • Scenes with a real person keep the rights attestation on record, and the action is audited
  2. 2026-11-18

    • Create a reusable character from a prompt and thirteen selections — style, age, hair, outfit, framing, light, palette and more — with an optional AI prompt enhancement applied once and reused for every shot
    • Render a full sheet of twelve emotions and twelve poses; shots render one at a time so the tool works within any plan’s concurrent-job limit
    • Place the same character into an image from another app: the scene is the source and the character rides along as the second reference
    • Import a character from a photo or an image URL — the vision model reads the appearance into a structured profile and the selections, with the rights attestation recorded
    • Real-person imports carry a likeness badge downstream; a report puts the character into the moderation queue and a reject removes it and its pending shots
    • Every render is a normal generation: per-shot pricing at the model’s real rate, the same plan/kill-switch gates, and character lineage on the generation
  3. 2026-11-18

    • Build a location plate from a prompt and selections — setting, interior/exterior, light, camera angle, lens, atmosphere, look, crowd, platform frame and up to eight set-dressing objects with placement and scale
    • A seed is stored on the location, so a take can be reproduced exactly and re-generated with only the changes you asked for
    • Modify a rendered plate with image-to-image: add or remove objects while the place, light and shot stay locked
    • Animate any plate with an image-to-video model (own model section, camera-motion presets, duration and cost)
    • Every change is an immutable version: append keeps earlier plates, replace swaps the set, restores copy a snapshot, and the full history is kept forever
    • Render a full lights × camera-angles coverage set (up to 20 labelled plates) that queues within your plan’s concurrent-job limit
    • Start from any library image: match its structure, or read the setting, light and angle back into the selections
    • Place a subject from another generation into any rendered plate, with perspective, horizon and light direction matched
    • Platform frames for YouTube, Instagram feed, Reels, TikTok, X and Facebook — only the ones the selected model can render
    • Optional prompt enhancement that rewrites the plate prompt without changing a single selection
    • Every set saves as a location with a stable id: reopen it, rename it, or delete it from the history panel
    • Attach a saved location to any generation as a reference image (inputs.locationIds), so a place can be reused across tools
  4. 2026-11-16

    • Blocklists from four public disposable-email feeds are merged and refreshed weekly instead of relying on one list
    • Sign-in, workspace invites and password resets now reject disposable addresses, not just signups
    • A brand-new throwaway domain is detected through its mail server even before it appears on a list

October 2026

10 updates
  1. 2026-10-21

    • Upload a reference image or video and free.flow analyses it into a reusable, runnable template
    • Analysis is metered on real vision tokens, so the cost is visible before it runs
    • Templates stay drafts until they pass validation and moderation, then activate for the workspace
    • Built templates join the existing template library and run like any other flow
  2. 2026-10-20

    • free.podcast: an optional music bed with volume and ducking, plus an outro sting, mixed by a new audio_mix mode
    • Re-render an episode from any line instead of starting the whole conversation over
    • Episode transcripts download as SRT
    • free.audiobook: chapter markers in the transcript, narration pace control and a transcript sidecar
    • Cloned voices now appear in both the podcast and audiobook voice pickers
  3. 2026-10-20

    • Text from one step flows into the slots of later steps, with the master input available everywhere
    • Fan-in steps (video join, production render, audio concat) take an ordered list of upstream inputs
    • Pause a run and resume it later, or retry a single failed step
    • Cleaner flow validation keeps stale steps from blocking a run
  4. 2026-10-20

    • Save and apply prompt presets from a strip in the generator, with style chips on the generate workspace
    • free.thumbnail gains image overlays (stickers and photos) composited server-side, plus per-layer weight, opacity and text outline
    • Daily login credits are published — the amount and expiry show on the site and in the dashboard
    • A first-run tour walks new workspaces through the generate page
  5. 2026-10-19

    • Brush a region mask directly on the image in free.edit and free.fill
    • Video: reframe the output, trim reel clips, tune real effect parameters and add a watermark
    • Provider-native music, sound-effect and speech controls are exposed instead of a shared subset
    • Audio mixer loudness targets and fades, with a transcript sidecar for long audio
    • Boards can be renamed, archived and restored, and the library downloads assets in bulk
  6. 2026-10-18

    • One-time direct payment per job from chat and the studios — no subscription needed
    • Dedicated pay-as-you-go and pay-per-generation pages, with a side-by-side comparison of the payment modes
    • Mobile quick top-up that lands on a prefilled wallet deposit
    • Video bills per resolution tier and speech per character, so quotes track provider cost
  7. 2026-10-14

    • Live support chat with one open conversation per user; replies arrive in realtime on both sides
    • Chat completions stream over WebSocket with an SSE fallback, and conversations are persisted
    • Accounts created through a social login can set a first password with an emailed one-time code
    • Account deletion asks for a short reason and offers two levels of cancellation feedback
  8. 2026-10-13

    • Each scene seeds its illustration from the last frame of the previous clip, so characters and sets stay consistent
    • Produce scenes one at a time and keep finished renders when a later scene fails
    • A per-scene motion prompt drives the animation step, and the audio-first pipeline times the edit to the narration
  9. 2026-10-05

    • Two new modes transcribe audio and video into text, with a language hint and vocabulary context
    • Read the transcript in the dashboard and copy it in one click, or fetch it from the API
    • Billed per input second, including mixed models that also generate text
    • Per-model input file and context limits are enforced from the catalog before a job starts
  10. 2026-10-05

    • OpenCode Go added as a provider in the model catalog
    • The generation MCP server authenticates per caller with a FreeInd API key and drives every local-media mode
    • The OpenCode chat integration syncs the model catalog automatically and supports every wire protocol
    • Benchmarks and examples share one paginated showcase grid

September 2026

17 updates
  1. 2026-09-30

    • A studio for original kids stories and nursery rhymes: script, scenes, narration and an assembled episode
    • Curated model presets, optional burned-in lyrics, sing-along rhymes and per-scene retakes
    • Viewers can submit story ideas and watch published episodes in the public gallery
  2. 2026-09-23

    • New free.edit studio: upload or pick a photo, describe the change, and run it on image editing models only
    • One-tap edit presets — background removal and replacement, restyle, product cleanup, color grade, object removal, relighting and upscaling
    • Model-driven controls: edit strength, aspect ratio, resolution, multiple results, seed and negative prompt
    • AI prompt enhancement rewrites your instruction into a precise editing prompt (+5 credits, paid plans)
    • Before/after compare slider, download, and an "edit again" loop that feeds the result back in as the next source
    • Included with paid plans; image editing is gated per workspace and can be paused platform-wide by staff
  3. 2026-09-23

    • New free.speech workspace: turn a script into a natural voiceover with speech models only
    • AI script writer: draft a spoken script from a brief, tone, audience and target length
    • Spoken-text prompt enhancer: expands numbers and abbreviations and tunes pacing (paid plans)
    • Voice cloning from a short reference sample where the speech model supports it
    • Voice, speed, format and delivery controls, with waveform player and MP3/WAV/OGG download
    • Public model catalog now distinguishes speech and music audio models
  4. 2026-09-23

    • Support ticket replies now appear live on both sides — no page refresh needed
    • An expanded help center backs faster, more consistent answers
    • Tickets are triaged by category, with abuse and security reports prioritized
  5. 2026-09-22

    • Renewal invoices now include a Cashfree payment link for one-tap renewal (India)
    • Checkout sessions carry through to the Cashfree client correctly
    • Payment events are applied only after signature verification, and metadata merges atomically so duplicate renewals cannot double-credit
  6. 2026-09-22

    • Usage charts: credits and spend over time, broken down by model and status
    • Wallet deposit history with amounts and status
    • Sign-in history so you can review recent account access
    • API key usage series and a storage mix chart
  7. 2026-09-21

    • New free.music studio: turn an idea into a full song with AI-written lyrics and music models
    • AI lyric writer with structure, mood and genre controls
    • Music prompt enhancement and an audio-type filter in the model catalog
    • Preview, download and manage songs from the dashboard library
  8. 2026-09-21

    • Storyboard editor: plan scenes on a timeline strip and generate each shot
    • Provider routing so every scene runs on the model you choose
    • Transition browser and basic editing tools for assembled reels
    • Voice samples pulled from the real text-to-audio collection
  9. 2026-09-21

    • Deleting your account now starts a grace period instead of an immediate wipe
    • Support can restore an account during the window; data is purged after it ends
  10. 2026-09-20

    • Cleaner model cards with single-row headers and provider tags
    • Beta badges for free.studio and free.flow
    • Consistent dashboard page width across the app
  11. 2026-09-16

    • Razorpay and Cashfree checkouts alongside Creem, with a live USD → INR rate for Indian customers
    • New mini and growth plans, plus all-models access on paid plans
    • Payment gateways can be enabled or paused individually without a deploy
    • Razorpay wired into billing reconciliation so renewals and refunds stay in sync
  12. 2026-09-15

    • Twelve ready-made free.flow templates across common use cases, seeded automatically
    • Full-screen model picker with provider pills and curated picks
    • Workflow steps validate prompts against each model's limits before running
  13. 2026-09-12

    • Rewards checkpoints: complete actions like joining Discord, following on X or upvoting on Product Hunt to earn credits
    • Referral links with pending rewards and celebratory emails when friends join
    • Affiliate tiers and promo codes
    • Public rewards page with the checkpoint timeline and tier ladder, served from live data
  14. 2026-09-12

    • Signups from disposable email domains are rejected to keep the platform free of throwaway abuse
  15. 2026-09-08

    • Quick-create tiles for every generation mode
    • Credit balance hero with plan and top-up shortcuts
    • Recent work thumbnail grid so you can pick up where you left off
  16. 2026-09-06

    • $5 Zero plan with a curated set of famous models free to use
    • Trial Pack: a one-time credit pack with 7-day expiry and a weekly purchase limit
    • Subscriber-free models: a rotating set of models free on paid plans
  17. 2026-09-04

    • Projects: group generations into projects and filter your library by project
    • Two-factor authentication (TOTP) from account settings
    • Usage breakdown by model and status

August 2026

23 updates
  1. 2026-08-26

    • Text-to-text generation alongside image, video and audio
    • Text output and prompts returned in generation responses and shown in the dashboard
    • MCP generation tools updated to support text models
  2. 2026-08-25

    • Markdown rendering for code blocks and formatted answers
    • Tool results stream into the conversation as they happen
    • Rich model and generation suggestion cards inside chat
  3. 2026-08-25

    • Search your asset library
    • Bulk favorite and delete
    • Generation info on each asset tile
  4. 2026-08-25

    • llms.txt and llms-full.txt routes describe the platform, models and docs for AI assistants and answer engines
  5. 2026-08-23

    • Public collections now live at readable /collections/<name>-<id> URLs
    • Shareable collection pages for same-prompt model comparisons
  6. 2026-08-23

    • Blue-green deploys: every release boots on an idle side, passes health checks, then takes traffic in a single atomic switch
    • Sitemaps render per request from live data instead of build-time snapshots
    • Failed deploys alert the team automatically and never leave the site degraded
  7. 2026-08-20

    • Video joiner: combine up to 10 clips into one video (paid plans)
    • Audio mixer: replace or mix the audio track on a video (paid plans)
    • AI prompt enhancement for music models, plus an audio-type filter in the catalog
  8. 2026-08-19

    • WhatsApp-style chat workspace with persistent sessions
    • File uploads and slash commands
    • Model switching and generation tools inside the conversation
  9. 2026-08-15

    • OpenAI-compatible /v1/openai/* routes for OpenCode-compatible clients
    • MCP servers for generation operations and gateway access
    • Admin status page for gateway health
  10. 2026-08-14

    • Submit support tickets from the dashboard and track their status
    • Categories for general, billing, abuse, security and account issues, with abuse and security reports prioritized
    • Internal notes and first-response tracking for faster resolutions
  11. 2026-08-12

    • Chain generation steps into reusable multi-step workflows
    • Editor with per-step models, prompts and schema validation
    • Run workflows end to end with cost estimates before you start
  12. 2026-08-12

    • New dark-first design across the public site and dashboard
    • Rebuilt homepage with modality sections and a capabilities grid
    • FAQ page with search and accordions
  13. 2026-08-10

    • Quickstart, authentication, examples, webhooks and billing guides
    • UI guide with device- and theme-aware screenshots and GIFs
    • OpenAPI reference published from the API
  14. 2026-08-10

    • Credit price standardized at $0.01, with per-generation costs shown before you run
    • Admin cost and margin views so pricing stays auditable
    • Provider spend tracked per model and per workspace
  15. 2026-08-09

    • Maker pages for OpenAI, Google, Black Forest Labs, ByteDance and more
    • Sample galleries and free-tier pricing on model cards
    • Dynamic sitemap covering maker and model pages
  16. 2026-08-08

    • Replicate, Together AI, Fireworks AI and DeepInfra integrated with normalized catalogs
    • Hundreds of image, video and audio models imported with pricing coverage
    • Same model across providers: compare rates and pick where it runs
  17. 2026-08-07

    • Vercel AI Gateway and fal.ai integrated with live catalog sync
    • Maker branding: models are shown by their original maker, never the routing provider
    • Provider logos and brand marks across the catalog
  18. 2026-08-03

    • New text-to-audio generation mode: scripts to natural narration and voiceovers
    • Voice, speech-rate (0.5–2×), and MP3 / WAV / OGG format control
    • Audio assets with waveform previews in the dashboard library
    • Audio models in the catalog with per-generation credit pricing
    • Admin model editor and generations filter support the audio modality
  19. 2026-08-03

    • Content moderation queue: automatic review of generations with allow, blur or reject decisions
    • SSRF protection on provider output downloads and CSRF checks on state-changing requests
    • CSP, HSTS, referrer and permissions-policy headers across the web app
    • Restricted-region checkout gate and audited staff impersonation controls
  20. 2026-08-02

    • Platform foundation: API, web app, background workers, Postgres and queues
    • Email and social sign-in with required email verification and password reset
    • Workspaces with team members, roles and email invitations
    • Public website with docs, trust pages and SEO
  21. 2026-08-02

    • Generation pipeline for text-to-image, text-to-video, image-to-image, image-to-video and video-to-video
    • Immutable credits ledger with per-generation reservations and transparent costs
    • Creem subscriptions with verified webhooks; pay-as-you-go wallet with a $10 minimum
    • Media processing: metadata stripping, thumbnails, video posters and retention
  22. 2026-08-02

    • Generation workspace for all launch modes with model-aware controls
    • History, detail views with event timelines, and an asset library
    • Billing center: plans, wallet and payment history
    • Discord and Telegram notifications
  23. 2026-08-02

    • Scoped API keys with rate limits, Idempotency-Key support and usage reporting
    • Signed customer webhooks with retries and delivery history
    • Admin console: users, workspaces, generations, pricing approvals, audit log and impersonation

Cookies keep FreeInd working. Optional analytics require your permission; ads for free accounts are served by Google, which shows its own consent message where required. Read our cookie policy.