MODEL CATALOG / PRIVATE PREVIEW
Choose the engine.
One request format. Different capabilities. Start with the smallest model that meets your workload.
MODEL / USE CASESIZECONTEXTINPUT PRICE
↗
8B32K$0.12/ 1M input tokens↗Stride 8B InstructEveryday chat, extraction and tool calls.
◇
32B128K$0.65/ 1M input tokens↗Summit 32B ReasonMulti-step analysis and long document synthesis.
⌘
14B64K$0.28/ 1M input tokens↗Forge 14B CodeCode completion, patches and repository questions.
∴
768 dims8K$0.02/ 1M input tokens↗Embed 768Compact document vectors for semantic retrieval.
∷
1,536 dims16K$0.05/ 1M input tokens↗Embed 1536Multilingual retrieval with a larger vector space.
≋
JSONL50K rows50%off interactive rates↗Batch processingAsynchronous processing for offline workloads.
No model matches that search. Try a task, model family or size.
6 catalog entries · Prices in USD · Model IDs and limits are preview specifications.
Need predictable capacity?
Dedicated endpoints reserve a model replica for your project. Send your token volume, peak concurrency and region preference.
Request an endpoint →Bring your existing client.
Set a base URL and model ID in an OpenAI-compatible client. Request and response schemas stay familiar.
See a request →