Models
| # | Model | Intelligence Index | Price / 1M |
|---|---|---|---|
| #1 | claude-opus-4-8 claude-opus-4-8 | 55.7 | $5.00in·$25.00out |
| #2 | gpt-5.5 gpt-5.5 | 54.8 |
| # | Model | Intelligence Index | Price / 1M |
|---|---|---|---|
| #1 | claude-opus-4-8 claude-opus-4-8 | 55.7 | $5.00in·$25.00out |
| #2 | gpt-5.5 gpt-5.5 | 54.8 |
| #3 | claude-opus-4-7 claude-opus-4-7 | 53.5 | $5.00in·$25.00out |
| #4 | Z.ai: GLM 5.2 glm-5.2 | 51.1 | $1.40in·$4.40out |
| #5 | gemini-3.5-flash gemini-3.5-flash | 50.2 | $1.50in·$9.00out |
| #6 | claude-sonnet-4-6 claude-sonnet-4-6 | 47.2 | $3.00in·$15.00out |
| #7 | gemini-3.1-pro-preview gemini-3.1-pro-preview | 46.5 | $2.00in·$12.00out |
| #8 | Qwen3.7 Max qwen3.7-max | 46.0 | $2.50in·$7.50out |
| #9 | Deepseek V4 Pro deepseek-v4-pro | 44.3 | $0.43in·$0.87out |
| #10 | claude-opus-4-6 claude-opus-4-6 | 43.7 | $5.00in·$25.00out |
| #11 | Deepseek V4 Flash deepseek-v4-flash | 40.3 | Free tier$0.14in·$0.28out |
| #12 | GPT-4o mini gpt-4o-mini | 6.9 | $0.15in·$0.60out |
Intelligence Index from Artificial Analysis — best-effort score per model, refreshed daily.
Meta-brand
th-orchestraPlan → build → review. Each request is routed to one of three model pools — a planner for architecture, one pinned executor for the build, and a review pass. Image requests always go to a vision-capable model. You pay for the model that actually answered — no Orchestra surcharge.
Within a session the build stays pinned to one executor model unless a clearly different stage (plan or review) is detected — so the upstream prompt cache stays warm instead of cold-starting each turn.