18 text. 27 image. 71 voices. One key.

Flagship reasoning model with vision, tool calling, deep research, and multimodal capabilities.

Frontier multimodal model. Advanced reasoning, analysis, and general purpose intelligence.

Fast conversational model with low latency. Ideal for chat, assistants, and customer support.

Highest throughput model for batch processing. Handles long contexts efficiently.

Tuned for creative writing, marketing copy, and brand voice consistency.

Multilingual model with native support for English, Zulu, Xhosa, Afrikaans and 50+ languages.

Hardened agentic model for autonomous task execution and tool chaining.

Compact, distilled model. Edge-deployable, 90% of ALM quality at 10% of the cost.

Precision code generation. Optimized for writing, debugging, and refactoring code.

Lightweight conversational model for high volume assistants and bots.

Deep research model. Long horizon reasoning with citation aware synthesis.

Fast high quality image generation. Photorealistic, artistic, anime, and abstract styles.

Stable Diffusion XL through our unified image endpoint. Versatile, fast, OpenAI compatible.

Photorealistic flagship at 1024px and beyond. Heavy detail, cinematic lighting.

Stylised painterly imagery. Editorial illustration, concept art, anime, watercolour.

Brand tuned product photography and editorial realism. R0.60 per image.

Warm female voice, the default Agentics persona. Streaming PCM 44.1kHz.

Clinical female voice. Fast, precise, ideal for technical or instructional content.

Deep, calm male voice. Authoritative tone for narration and longform content.

Playful, conversational male voice. The friend you wish you had on the line.

Speech to text with VAD, end-of-turn detection, and language autodetect.

GTE-Small embeddings. 384 dimensions. Batch up to 1024 inputs per call.