AI MODELS IN ASTRA / CHAT, IMAGE, VIDEO & VOICE

The next model.
Your next possibility.

Claude, Gemini, GPT, Seedance, MiniMax, FLUX and more. Choose your model inside Astra, then carry the idea into your agents, creative projects and finished work.

Request access

CATALOG CHECKED 14 NOVEMBER 2026

FIND THE RIGHT FIT / POWERED BY OPENROUTER DATA

Different strengths.
Same place to create.

Choose two models and explore what changes. Start with the inputs you have, the format you need and the controls that matter to your work.

Claude Fable 5.2

Input types
Text, image, file
Context window (provider)
1,000,000 tokens
Tool calling
Listed
Structured outputs
Listed
Explore in Astra OpenRouter model details ↗

Gemini 4 Hydrogen

Input types
Audio, file, image, text, video
Context window (provider)
1,048,576 tokens
Tool calling
Listed
Structured outputs
Listed
Explore in Astra OpenRouter model details ↗

Provider specifications checked 14 November 2026. These are OpenRouter capabilities, not Astra plan limits or measured output-quality results. Available controls depend on the connection and model selected in Astra.

Explore OpenRouter’s model comparison ↗

ONE PROMPT / TWO MODELS / MADE IN ASTRA

One brief.
Two interpretations.

The same samurai reference. The same prompt. Seedream 5.5 and Nano Banana 2.1, both created through Astra’s OpenRouter workflow.

The last snowfallCharactersMADE IN ASTRA
THE STARTING POINT

Original Astra character reference.
Both requests: 4:3 canvas, 2K resolution.

Seedream 5.5

The samurai / Seedream studyCharacter sheetsMADE IN ASTRA

Nano Banana 2.1

The samurai / Nano Banana studyCharacter sheetsMADE IN ASTRA
See the shared prompt

Create a premium realistic character design sheet using the attached original samurai reference. Keep the same adult face, tied-back dark hair, dark indigo lamellar armor, muted red ties, fabric layers and proportions. On one clean warm-gray landscape canvas, arrange three evenly spaced full-body views: front, true side profile and back. Add one smaller detailed three-quarter face portrait in the upper right without overlapping the figures. The same person, costume, materials and equipment must remain consistent across all views. Photorealistic film costume-development quality, finely resolved fabric, worn lacquer and realistic skin, soft neutral studio light, grounded believable anatomy, hands relaxed at his sides, sheathed swords only. Elegant spacious art-department presentation. No labels, lettering, logos, watermark, frame, scenery, extra characters or duplicate limbs.

A single creative study, not a benchmark. Models interpret a brief differently and results vary between generations.

Bring your own idea

ACTUAL ASTRA OUTPUTS / SAME BRIEF

See the difference.
In motion.

One opening frame, one prompt, two fifteen-second interpretations. Compare how local MiniMax H3 and cloud Seedance 2.5 stage the same sword fight.

Explore both action scenes
A SAMURAI ACTION STUDY

The last gate

LOCAL GENERATIONMiniMax H3
15.1s · 864 × 480
Two samurai locked in a close sword exchange before a snowy gate
Two samurai trade close sword strikes in a snowy courtyard. One lunges and the other blocks as they stay in front of the lantern-lit gate.
CLOUD GENERATIONSeedance 2.5
15.0s · 1280 × 720
Samurai crossing swords in a snowy courtyard beside a row of lanterns
Two samurai advance across a snowy courtyard framed by lanterns. Their blades meet in bright sparks. They withdraw to either side of the gate with their swords ready.

Same opening frame. Same prompt.
Two actual outputs made in Astra.

See the shared prompt

Animate the supplied first frame. Preserve both fighters, their identities, costumes, the location and the visual style throughout. Create a premium photorealistic historical action-film scene, fifteen seconds of continuous, clearly readable sword-fight choreography. Snowy stone courtyard outside a red temple gate, warm lanterns, deep blue winter dusk. Exactly two adult fighters: the reference samurai and a distinct opponent in plain charcoal armor and a closed helmet. Start in a medium-wide full-body two-shot with both already moving, not a portrait. Seconds 0–4: the opponent advances with two fast slashes; the samurai blocks each and steps diagonally across the wet stones. Seconds 4–10: they trade a fluid sequence of controlled blade clashes, parries and a low duck, traveling sideways through the courtyard; snow sprays under their feet and tiny sparks appear only where the steel meets. Seconds 10–15: the samurai turns past the next strike, knocks the opposing blade aside, and both recover into balanced opposing stances as falling snow fills the space between them. Smooth low lateral tracking camera, one continuous shot, normal speed, precise grounded footwork, clear silhouettes, visible hands holding solid swords, consistent anatomy and costume, physically coherent contact and momentum. Rich tactile materials, restrained cinematic contrast, realistic breath in cold air. No long static pose, no slow motion, no frantic cuts, no teleporting, no extra fighters, no duplicate limbs, no blade passing through a body, no blood, injury or gore. No text, logos, dialogue or music. Subtle wind and steel sounds only.

Individual Astra generations, with native output dimensions shown beside each film. Both website edits are silent.

18 selected models

CHAT

Think it through.

IMAGE

Find a look worth keeping.

VIDEO

Put the world in motion.

Models shown were present in Astra’s selectable catalog and checked against OpenRouter. These are starting points for exploration, not a universal quality ranking. “Popular” refers to OpenRouter usage. Supported inputs, tools, output settings, access and token use vary by model and account.

ELEVEN V4 / ORIGINAL ASTRA CREATION

A whisper.
A little wonder.
A whole new voice.

Direction changes the feeling. Hear one voice move through curiosity, a breath, a laugh and a moment of excitement.

[whispers][curious][laughs][excited]
SARAH / ELEVEN V4CREATED IN ASTRA
Explore the tagged script

[whispers] Come closer. Every world starts with a sound.

[curious] Is it a voice on the other end of the line... or a character stepping out of your imagination?

[exhales] There it is.

[mischievously] Oh, this is going to be good.

[laughs]

[excited] Roll the camera. Build the scene. Make something nobody has heard before.

[applause]

[warmly] A thought. A conversation. A world you have not seen yet. Welcome to Astra.

ElevenLabs audio-tag guide ↗

AI-generated creative narration. This sample demonstrates the voice model; it is not a recorded Phone Agent call.

THE MODEL IS THE BEGINNING

Keep the idea moving.

Start in a conversation.

Explore the brief with a chat model and the project context you choose.

Give it a visual world.

Make images, develop video shots and keep selected takes in your Library.

Bring it to a finish.

Build the sequence, add voice and music, refine the timing and export your edit.

YOUR WORK. YOUR IDEAS. YOUR ASTRA.

Your next chapter.
Starts here.

A plan built for makers. Usage across every feature. $99 per month per user.

Request accessExplore the $99 package · Request a demo

Astra product interface

Genuine earlier app screenshot · Sample data

Selected work from Astra