Skip to main content
Glama

Generate AI Video

antix_generate_video

Generate 1–4 independent AI videos from one prompt, optionally with a starting image. Set count to the requested amount (default 1); all videos queue atomically and run in parallel, while each has its own task, charge receipt, and failure refund. duration is any whole second the model renders, priced per second: kling-v3/kling-v3-fast 3-15, omni-flash 3-10, seedance 4-15, seedance-2-5 4-30, h3-max 5-15, veo exactly 4/6/8 (another value snaps to the nearest length the model renders). The response preserves taskId for the first output and also returns batchId, taskIds, and tasks; for count > 1 poll once with antix_check_batch, otherwise call antix_check_task (typically 30–90s). Supports character/scene continuity from a starting frame. Charged at the same rate as the AIGE web app. Unsure which model? Call antix_pick_model (free, instant).

Models (credits at 5s/8s/10s): • omni-flash — 50/80/100 cr at the default 720p, audio BUNDLED (dialogue, sfx, ambience in the same pass). Priced per second BY resolution: 360p is 15/24/30 cr (the cheapest clip with sound anywhere in this list), 1080p is 75/120/150, 4k is 150/240/300 — so state the tier when you quote it. Supports text/image/reference-to-video, an optional endImage, and prompt-based source-video editing (editing is a flat 140 cr at 720p). First pick for fast talking clips and social content with sound. • kling-v3 — 84/135/168 cr with audio (the default), 56/90/112 cr silent (audio: false). Top cinematic quality, fluid character animation, start+end frame control. • kling-v3-fast — Kling V3 Standard, the lower-cost tier: 63/101/126 cr with audio, 42/68/84 cr silent. Pick when cost or turnaround matters more than polish. • veo — 240/320/320 cr, audio included. Google's flagship: best prompt adherence and physical realism; renders exactly 4/6/8s (5→6s, 10→8s). Retiring: Google retires Veo 3.1 on 17 Nov 2026; once AIGE switches it off the call is refused before any charge — use omni-flash instead (1080p costs what Veo Fast did). • seedance — 152/242/303 cr, audio included. Motion-rich dynamic action, expressive camera moves, start+end frames, and multimodal reference mode. • seedance-2-5 — 232/370/463 cr, audio included. The longest single take: any whole duration from 4 to 30 seconds as one continuous shot, priced per second (30s = 1387 cr at 720p, 617 cr at 480p — long takes are expensive, so confirm the length with the user before queueing one). Caps at 720p. Takes up to 50 references (30 images / 10 videos / 10 audio) to lock a character, set and palette across the whole shot. Pick it for long single-shot storytelling; pick seedance when 1080p matters more than length. • h3-max — 40/64/80 cr at 768P, 25/40/50 cr at 480P: the CHEAPEST video in the catalogue, with audio BUNDLED at no extra cost. Audio is always on and cannot be disabled (passing audio: false is rejected rather than silently ignored). Any whole duration from 5 to 15 seconds, priced per second. Supports text/image-to-video plus an optional endImage for a directed transition. Pick it for bulk drafts, animatics, b-roll and silent loops.

Starting-frame priority: pass sourceImage (HTTPS URL or base64 data URL) when the user attaches/provides a specific frame. Legacy sourceImageUrl also works. If there is no explicit source image but the user named a saved AIGE character, pass characterId — the server will use that character's latest gallery image, then avatar, before falling back to text-to-video. endImage / endImageUrl can provide a tail frame for models that support it.

Video/audio sources from chat: sourceVideo, referenceVideos and referenceAudios accept HTTPS URLs only — video bytes are too large for base64 tool args. The user path: upload at https://aige.ws/r (accepts MP4/WebM/MOV ≤50MB and MP3/WAV/OGG/M4A/FLAC ≤15MB, returns a short link) → paste the link → you forward it in the matching field. A URL of a previous AIGE generation (from antix_get_history or a completed task) works directly. If the user attached a video in chat and you cannot access it as a URL, do NOT silently generate without it — tell them: 'Upload it at https://aige.ws/r and paste the short link here.' Then wait.

Input Schema

NameRequiredDescriptionDefault
audioNo
countNo
modelNo
promptYes
durationNo
endImageNo
messageIdNoOptional originating chat message id for inline media reconciliation.
projectIdNoAIGE work-project id that should receive pending and completed assets.
sessionIdNoOptional AIGE agent session id within projectId.
resolutionNo
aspectRatioNo
characterIdNo
endImageUrlNo
sourceImageNo
sourceVideoNo
idempotencyKeyNoStable client-generated key. Retrying the same request with the same key must not charge or queue twice.
sourceImageUrlNo
referenceAudiosNo
referenceImagesNo
referenceVideosNo

Schema Changelog

Changes observed during successful MCP inspections.

  1. First observed

TDQS

A4.4/5.0
Behavior3/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

The annotations only declare destructiveHint=false and idempotentHint=false, so the description carries the behavioral burden and does add real value: atomic queueing, per-video charge receipts and failure refunds, ~30–90s polling latency, forced-audio rejection on h3-max, and the Veo retirement cutoff. However, most of the volume is pricing/marketing detail rather than operational behavior (no mention of concurrency limits, what happens to partial batch failures, or how long assets persist), so it is strong but not exhaustive.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness3/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Front-loaded correctly: the first two sentences cover the core action, batching, and polling. But the body balloons into a multi-tier credit price list per model and per resolution, which is far more than an agent needs to select and invoke the tool and buries the operational rules (source-image priority, upload path) mid-document.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 20-parameter, 7-model generation tool with no output schema, the description explains the return shape (taskId, batchId, taskIds, tasks), the correct follow-up polling path per count, billing/refund semantics, and input-source prerequisites. An agent has everything needed to call this correctly.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters5/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

With only 20% schema description coverage across 20 parameters, the description compensates heavily: it constrains `duration` per model (including Veo's exact 4/6/8 snap behavior), documents count semantics, sourceImage vs legacy sourceImageUrl vs characterId priority, endImage as tail frame, HTTPS-only rules for sourceVideo/referenceVideos/referenceAudios, and resolution-dependent pricing. Only aspectRatio and the projectId/sessionId plumbing go unaddressed.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Opening sentence states a specific verb and resource with scope: 'Generate 1–4 independent AI videos from one prompt, optionally with a starting image.' It also distinguishes itself from siblings by naming follow-up tools (antix_check_batch vs antix_check_task) and by routing model uncertainty to antix_pick_model.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives explicit when-to-use rules per model ('First pick for fast talking clips and social content with sound', 'Pick it for bulk drafts, animatics, b-roll', 'pick seedance when 1080p matters more than length'), plus when to use batch polling, how to choose sourceImage vs characterId, and a hard when-not with the exact user-facing message for inaccessible chat videos.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

Try in Browser

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources