Release of Gemini 3.7 Flash, Grok 4.6, Muse Spark 1.2, and pricing changes!
Steven · 2026-08-13
Price cuts, new models, and a couple of overdue fixes.
FEATGoogle cut gemini-3.6-flash in half, $1.50 to $0.75 input and $7.50 to $3.75 output, cached input and storage following the same ratio. It's introductory pricing through 2026-12-31, stepping back up to the original rate on 2027-01-01.
FEATAdded Gemini 3.7 Flash (gemini-3.7-flash), released today at the same $0.75/$3.75 rate and the same introductory window as 3.6 Flash.
FEATGoogle also flagged gemini-2.5-flash for retirement, shutting down 2027-05-07 in favor of gemini-3.5-flash-lite.
FEATMeta added Muse Spark 1.2 (muse-spark-1.2), priced the same as 1.1: $1.25 input, $0.15 cached input, $4.25 output.
FEATAlongside it, a Muse Spark 1.2 Contributor tier (muse-spark-1.2-contributor), the same model at $0.10 input and $0.20 output for opting in to Meta training on your traffic.
FEATSpaceXAI's grok-4.5 cached input dropped from $1.00 to $0.60 per million on the >200K tier.
FEATAdded Grok 4.6 (grok-4.6), released yesterday, priced the same as 4.5: $2 input, $6 output.
FIXgrok-3 was still marked deprecated. It's retired.
FIXBackfilled claude-opus-4 as a retired model, missing from the table entirely until now.