Kiro
Kiro AI IDE — AWS's spec-driven agentic coding client, the official successor to Amazon Q Developer
Core models
The latest Opus in Kiro's catalog with 1M context at a 2.2× credit multiplier vs Auto; strongest self-verification—flags uncertainty instead of false progress—suited to high-stakes architecture and complex spec implementation.
Auto-scales reasoning depth by task complexity with precise instruction following and higher-resolution vision; 2.2× multiplier, Pro and above—suited to complex agent tasks needing adaptive deep thinking.
Stable benchmark performance, stays focused over long sessions, strong code review; 1M context at 2.2×, Pro and above—suited to extended debugging and multi-round spec iteration.
Strong on cross-system architecture and ambiguous boundaries with high single-shot accuracy; 200K context at 2.2×, Pro and above—suited to service-boundary analysis and architecture tradeoff work.
Near-Opus intelligence at Sonnet cost with better token efficiency; 1M context at 1.3×, Pro and above (not on Free)—the value pick for daily high-intelligence tasks.
Additional core model names still appear above, with full details on the latest official page.
Plan details
Quota
Notes
- Credits deduct fractionally by request complexity, not one credit per message; Auto is the 1.0× baseline, Opus 2.2×, Sonnet 1.3×, Haiku 0.4×, Qwen3 Coder Next 0.05×.
- Free includes Sonnet 4.5/4.0, Auto, and several open-weight models; all Opus variants, Sonnet 4.6, and Haiku 4.5 require Pro or above. First paid upgrade grants a one-time 1,000-credit ($20 equivalent) bonus.
- Kiro is the official successor to Amazon Q Developer, which stopped new signups and ends 2027-04-30. Enterprise pricing is via AWS sales, separate from individual IDE subscriptions.
Supported coding tools
Pricing and model data sourced from official vendor websites