OpenCode Go × Artificial Analysis

Unofficial Go $10/mo II v— Snapshot — Updated —
Cost per task at OpenCode Go rates · source: Artificial Analysis
Models (scored)—
Go models—
Plan price$10 / mo

Intelligence Index vs. Cost / Task — Pareto frontier

Discount
Labels
Vendors
Color = vendor · Scroll to zoom · Drag to pan

Models by cost / task

Orange = on the frontier
# Model Vendor Intelligence Index Cost / task
Go $/task = AA $/task × $10 / monthly limit; non-Go models at AA list price.
Muse Spark appears twice: Meta standard at AA list price, and Go's Contributor price for the same model and Intelligence Index. AA's score is the Max effort run; the Contributor variant Go sells cannot use Max.
Beam appears twice in Go price: AA list price and Go's free tier ($0/task, no monthly allowance, limited regions and limited time, daily request cap). Free models are handled differently: $0 cannot sit on a log scale, so every $0 row is drawn in its own FREE column left of the cheapest paid price, with a dashed rule marking where the paid scale starts. In the combined view, Original price shows only the AA list-price row, so the free tier is not listed twice.

Data sources — AA Data API (Intelligence Index, cost/task, list token prices) · OpenCode Go (monthly limits) · OpenCode Console models (actual token prices) · Go models API.
Discount method — OpenCode does not always label a discount; it often just lowers the price. We use AA's list price per 1M tokens as the reference — the most widely accepted public price today — and compute discount = 1 − OpenCode actual price / AA list price. Where AA list pricing is unavailable, the struck-through Console list price is used as the reference. This applies to Go models only and leaves the Original (list) price untouched. We keep this method unless a dispute requires a change.
Static snapshot (not live) · AA Intelligence Index · Go pricing · unofficial, not affiliated with OpenCode / SST.