Catalog refresh: 26 models across 10 labs

Ahead of launch we swept the catalog and finalized the launch lineup: 26 models across all ten labs, from trillion-parameter flagships to free flash tiers.

Highlights include DeepSeek-V3.2 for general chat and tools at $0.28 per 1M input tokens, GLM-4.6 for coding and agentic workflows, Qwen3-Max for maximum quality, and Kimi K2 for long-document agents.

If you’re prototyping on a zero budget, GLM-4.5 Flash, Doubao Seed 1.6 Flash and Spark Lite are completely free. Full context windows, capabilities and indicative prices are on the Models page.