All models
Image generationLive
QuiverAIArrow 2
QuiverAI's flagship model for turning prompts, images, and existing SVGs into editable vector graphics and micro-animations.
Plain-English overview
What Arrow 2 actually is
Arrow 2 generates SVG code instead of a flattened bitmap. Designers and developers can inspect and edit the resulting paths, fills, groups, and animation rather than treating the output as a finished rectangle of pixels.
The same model supports text-to-SVG creation, SVG editing, raster-to-vector conversion, reference-guided variations, and small animations. QuiverAI positions it as the everyday balance of speed, quality, and cost, while Arrow 2 Telos is the higher-priced option for longer briefs and more difficult assets.
Good fit for
- Editable icons, illustrations, and design-system assets
- Turning sketches or raster references into SVG
- Technical diagrams and lightweight web animation
Category comparison
The facts that matter for image models
These are provider-published specifications, not Cody benchmark scores. Follow the linked sources for current limits and endpoint-specific exceptions.
- Typical price basis
- $4 input · $20 output / 1M tokensA representative current provider price with its quality or resolution basis.
- Maximum output
- Resolution-independent editable SVGMaximum documented resolution or pixel area for the listed route.
- Core tasks
- Generate, edit, vectorize, and animate SVGWhether the model supports generation, editing, or both.
- Reference images
- Up to 14 reference imagesHow the model can use images to guide or edit an output.
- Text & layout
- Structured SVG output; test fonts, paths, and editable textProvider-documented positioning for typography and structured composition.
- API access
- Quiver app, direct API, and Vercel AI GatewayDirect and notable third-party routes included in this guide.
Pricing & comparisons
Estimate your cost
Set your usage. Your estimate updates as you type.
Example output: square 1,024 × 1,024 images. Prices depend on supported resolution and quality; edits and retries can cost extra.
Arrow 2
QuiverAI
Pricing & comparisons
Estimates exclude taxes, tools, cache storage/writes, free allowances and custom discounts. Image estimates cover output only, not prompt or reference-image charges. Quality modes differ by model. Unlisted settings are not treated as free.
How this estimate works
Estimates exclude taxes, tools, cache storage/writes, free allowances and custom discounts. Image estimates cover output only, not prompt or reference-image charges. Quality modes differ by model. Unlisted settings are not treated as free.
API and provider access
Where to get Arrow 2
QuiverAIArrow 2 model and pricingOfficial model ID, token rates, reference-image limit, and intended generation, editing, animation, and vectorization uses.QuiverAIRead the Arrow 2 announcementOfficial launch examples and QuiverAI's claims about geometry, composition, vectorization, and micro-animation.Vercel AI GatewayUse Arrow 2 through VercelGateway access under the quiverai/arrow-2 identifier; verify live provider and gateway terms before production.
Availability
Regions and access stage
Generally available in the Quiver app and API, with a Vercel AI Gateway route also documented.
QuiverAI's public model materials do not enumerate model-specific processing regions. Confirm deployment and contractual terms for regulated or residency-sensitive work.
Check live availabilityData and training
The route matters.
New QuiverAI organizations default to allowing storage and model training. Organization and project settings can restrict data to benchmark storage without training or forbid storage, with the strictest setting taking precedence. QuiverAI says SOC 2, ISO 27001, SSO/SAML, and SCIM are not currently available.
This is a concise reading of the cited provider material, not legal advice. A third-party gateway can have different storage, routing, training, and residency terms from the model maker's direct API.
Read the provider policyFrequently asked questions
Arrow 2 FAQ
What is Arrow 2?
QuiverAI's flagship model for turning prompts, images, and existing SVGs into editable vector graphics and micro-animations. Arrow 2 generates SVG code instead of a flattened bitmap. Designers and developers can inspect and edit the resulting paths, fills, groups, and animation rather than treating the output as a finished rectangle of pixels.
When was Arrow 2 released?
Arrow 2 was released on September 7, 2026 according to the cited provider materials.
Where can I access Arrow 2?
Generally available in the Quiver app and API, with a Vercel AI Gateway route also documented. The access routes listed in this guide are QuiverAI and Vercel AI Gateway.
How much does Arrow 2 cost?
$4 input · $20 output / 1M tokens. QuiverAI lists cached input at $0.40 and cache writes at $5 per million tokens. Because Arrow generates SVG as tokens, the cost of one asset varies with prompt length, reference material, and SVG complexity.
Where is Arrow 2 available?
Generally available in the Quiver app and API, with a Vercel AI Gateway route also documented. QuiverAI's public model materials do not enumerate model-specific processing regions. Confirm deployment and contractual terms for regulated or residency-sensitive work.
Is my Arrow 2 API data used for training?
New QuiverAI organizations default to allowing storage and model training. Organization and project settings can restrict data to benchmark storage without training or forbid storage, with the strictest setting taking precedence. QuiverAI says SOC 2, ISO 27001, SSO/SAML, and SCIM are not currently available. The policy belongs to the provider route and account terms, so verify it again before production use.
Related comparisons
Head-to-head comparisons
Arrow 2 vs GPT-Image-2.5 Flare
OpenAI's speed-first GPT Image 2.5 model for everyday generation and editing, positioned at GPT Image 2-level image quality with lower latency.
Open full comparisonArrow 2 vs Nano Banana 2
Google's versatile image workhorse for fast generation, conversational edits, multiple references, grounded imagery, and outputs up to 4K.
Open full comparison
