Plan event · 23 Sept 2026
All plan events · 31 providers tracked
CommandCodemodels
What moved
Go, GOAT, Pro, Max 10x, Max 20x and Provider served the StepFun family as Step 3.7 Flash and Step 3.5 Flash only → Step 5 Preview added to Go, GOAT, Pro, Max 10x, Max 20x and Provider at $1.00/$2.70 per 1M (cache read $0.05; per-model allowance $20 GOAT / $30 Pro)
CommandCode lists stepfun/Step-5-Preview ("600B sparse-MoE agentic coding with 1M context") in its CLI models page, and the pricing-limits page publishes the row ($1.00 input / $2.70 output / $0.05 cache read per 1M, no cache-write band, single rate tier). The availability map sets every tier true including individual-go (minPlanName "Go"), so the model lands on all six tracked plans. Per-model allowance is the standing GOAT $20 / Pro $30 shape — the announcement's "2x usage in GOAT" is not published anywhere and is not encoded.
Verified 2026-09-23 · source
Affected plans
More CommandCode events
- 2026-09-30 · Models — CommandCode now serves a second fast-tier DeepSeek row, DeepSeek V4.1 Flash Fast, on every
- 2026-09-30 · Models — CommandCode shipped GPT-6.1 Sol the day OpenAI launched it (changelog 1.71.0, Add GPT-6.1
- 2026-09-30 · Models — CommandCode now serves Claude Sonnet 5.5 from GOAT upward. The provider pricing-limits pag
- 2026-09-28 · Models — CommandCode serves four paid models on every plan, including the $1 Go tier, that no track