Models

Curated for coding

Every plan includes fast, affordable models. Pro unlocks the frontier — Claude, GPT and Gemini — the same models behind tools like Claude Code. Use auto and we pick, or name any model.

Included on every plan

fast & affordable
ModelBest forContextInput /1MOutput /1M
Mason Auto default
auto
Smart default — routes to the best-value model131k
DeepSeek V3
deepseek-v3
Fast, excellent general coding131k $0.20$0.80
DeepSeek R1
deepseek-reasoner
Deep reasoning for hard problems164k $0.70$2.50
Qwen3 Coder
qwen3-coder
Code-specialised, 1M context1M $0.22$1.80
Kimi K2
kimi-k2
Long context, agentic coding131k $0.57$2.30
GLM-4.6
glm-4.6
Strong all-rounder200k $0.43$1.75
GLM-5.2
glm-5.2
Latest Zhipu flagship, 1M context1M $0.93$3.00
Llama 3.3 70B
llama-3.3-70b
Open model, blazing-fast & low-cost131k $0.10$0.32

Pro plan

frontier models
ModelBest forContextInput /1MOutput /1M
Claude Sonnet 5 Pro
claude-sonnet
Best-in-class agentic coding1M $2.00$10.00
Claude Opus 4.8 Pro
claude-opus
Maximum capability, hardest tasks1M $5.00$25.00
Claude Haiku 4.5 Pro
claude-haiku
Fast, lower-cost frontier model200k $1.00$5.00
GPT-5.4 Pro
gpt-5
OpenAI flagship1M $2.50$15.00
Gemini 3.1 Pro Pro
gemini-pro
Google flagship, 1M context1M $2.00$12.00

Prices are indicative per 1M tokens. Frontier models cost more, so they draw down your monthly budget faster — auto keeps costs low by default.