Pick the right engine for the job (Claude Opus for reasoning, Flux Pro for stills, ElevenLabs for voice) without juggling vendor SDKs, separate logins, or per-provider billing pages.
application and data on EU (Hetzner) infrastructure
Loading models…
Which model when?
A quick map of the five capabilities, and the engines we route by default.
Text
Reasoning, writing, code.
Frontier text models for analysis, long-form writing, and coding. Pick by context window when work spans long documents, by speed when you draft in bulk.
Photorealism, stylised art, typography, and brand-safe commercial generation. Each engine specialises: Flux for fidelity, Ideogram for type, Firefly for licence-clean assets.
Vector representations for semantic search, RAG, and clustering. Pick by retrieval quality benchmarks and language coverage. Embeddings are foundation, not a feature you see.
No. AresGen routes prompts to upstream vendors — OpenAI, Anthropic, Google, DeepSeek, xAI, and 20+ more — through its MultiModel routing engine. Models are the vendors'; the workspace, billing, and team workflows are ours.
Can I bring my own provider keys?▾
Not today. Model access is bundled into your plan credits, so no provider accounts or key management are needed. If your team has a custom requirement, talk to sales about enterprise terms.
Which models are included on the Ultimate plan?▾
Ultimate unlocks the full catalog including flagship frontier models. Pro covers most production workloads; Lite covers everyday writing, drafting, and image generation. See the pricing matrix for the exact tier of every model.
What happens when a vendor changes or deprecates a model?▾
The vendor's change applies upstream. We update our catalog within a release cycle, mark deprecated models, and prefer current flagship versions in featured slots. Your workspace history stays intact.
Are the context windows and training cutoffs accurate?▾
Each spec on this page is taken from the vendor's published documentation. We surface vendor-reported figures as-is and do not benchmark them ourselves. Use the comparison view to read two or three models side-by-side.
Can I A/B test the same prompt across multiple models?▾
Yes. The playground lets you run one prompt against multiple selected models and compare outputs by cost, latency, and quality. Use the Compare drawer here to shortlist, then jump to the playground.
Try the routed catalog free today.
Spin up a workspace and try the full multi-model catalog without picking a plan. Bring your team in once you find the right blend of engines for your work.
