Skip to Content

Models

Text Models

Model NameSpecialisation & Notes
llama-4-maverick-17b-128e-instructMultimodal-capable model with OCR support; can read images and convert them into structured text
llama-4-scout-17b-16e-instructHigh-throughput instruction model optimized for fast inference and production chat use cases
llama-3.3-70b-instructEnterprise-grade assistant optimized for summarization, analysis, and large-scale production workloads
llama-3.1-8b-instructFast, cost-efficient model for general conversation, summarization, lightweight agents, and Q&A.
gpt-oss-20bOpen-source mid-scale model (20B); suitable for balanced workloads, agents, and cost-controlled deployments.
gpt-oss-120bLarge open-source model for enterprise and research workloads; supports deep reasoning and extended context.
deepseek-r1-distill-llama-70bEfficient reasoning model distilled from R1; balances speed with strong analytical performance.
qwen3-32bRobust multilingual and multi-turn reasoning model; strong instruction following and coding support
qwq-32bDialogue-focused reasoning model optimized for conversational depth and multilingual workloads.
gemma2-9b-itInstruction-tuned assistant optimized for task-oriented chat and structured responses.
kimi-k2-instruct-0905Large-context reasoning model with cache pricing; strong for long-form analysis and research workflows.

Image Models

Model NameSpecialisation & Notes
image-gen.pro-1.0Advanced image generation; high-res outputs, varied visual styles, image editing, creating new image from multiple images
image-gen.pro-1.5Advanced image generation with enhanced quality; ultra-high-res outputs, expanded visual style range, sophisticated image editing capabilities, advanced multi-image composition and blending

Video Models

Model NameSpecialisation & Notes
video-gen.pro-1.0Text-to-video & image-to-video; rich media & configurable durations
video-gen.pro-fast-1.0Text-to-video & image-to-video; rich media & configurable durations. Cheaper and faster version
video-gen.liteImage-to-video; rich media & configurable durations

Depreciated Models

ModelTypeDeprecated On
qwq-32bTextJan 11, 2026
gemma2-9b-itTextJan 11, 2026
llama3-70b-8192TextDec 26, 2025
llama3-8b-8192TextDec 26, 2025
deepseek-v3-0324-671bTextDec 26, 2025
deepseek-r1-671bTextDec 26, 2025
Last updated on