Which model for which task

Curated selection guide, derived from our catalog. Data: all_models.json.

Model	Size	Context	Strength	Best for
GLM-5.3	753B MoE	202K	Flagship reasoning	Plan, code, analyze
Kimi K3	1T MoE	256K	Long-doc analysis	Books, legal, research
DeepSeek-V4-Pro	671B MoE	164K	Balanced reasoning	Chat, agents, code
Qwen3-Coder-480B	480B MoE	256K	Code specialist	IDE, agents, refactors
Llama-3.3-70B	70B	128K	Reliable generalist	Chat, RAG, fine-tune base
Qwen3-8B	8B	128K	Fast economical	High-volume, drafts, evals

All models available at the catalog. Flat £0.00395/M tokens on all plans.