MicrosoftGeneralReleased 2 Jun 2026
MAI Thinking 1
Microsoft reasoning (preview). Microsoft’s in-house general reasoning model for enterprise tasks on Foundry.
No independent index yet
$0.28100K-in + 10K-out request
$2.00Input / 1M tokens
$8.00Output / 1M tokens
—Speed not measured
Watch for: Public preview. Azure lists $2 / $8 per 1M tokens (cached input $0.20) for the global deployment. No independent index yet; vendor comparisons use older Claude 4.6 models.
Pricing, limits, speed & capabilities
Pricing
Standard rates per 1M tokens$2.00 in · $0.20 cached · $8.00 out
Long-prompt thresholdSame rate across the window
Not published
None offered
Limits & modalities
256,000 tokens
Max outputNot published
Inputs → outputText → text
AccessMicrosoft Foundry public preview; MAI Playground.
Speed & capabilities
Output speedNot measured
Time to first answer (incl. thinking)Not measured
Tool callingNot published
Structured outputNot published
Computer useNot published
Sources: Microsoft · MAI Thinking 1 · Microsoft Azure · Foundry Models pricing . Checked 21 Nov 2026.