Bolt: Sonnet 4.6 Becomes the Default Model

Sonnet 4.6 Is Now Bolt's Default Model

Bolt has upgraded its default AI model to Claude Sonnet 4.6, replacing the previously default Sonnet 4.5. The change took effect during the week of April 6–10, 2026, and reflects Bolt's assessment that Sonnet 4.6 delivers a meaningfully better experience across the full range of everyday development tasks — from planning and coding to debugging and iterative refinement.

What Changed

Sonnet 4.6 is now the model Bolt selects automatically when users start a new project. According to Bolt's documentation, it "offers the best balance of speed, cost, and reasoning depth for everyday building tasks." The model brings deeper reasoning capabilities compared to its predecessor while remaining priced at a level suitable for extended, all-day use.

As part of this model lineup refresh, Sonnet 4.5 and Opus 4.5 are no longer available in Bolt. The streamlined model selection now offers three tiers:

• Haiku 4.5 — Fast and token-efficient, best suited for quick edits and simple tasks
• Sonnet 4.6 (new default) — Balanced performance for everyday development work
• Opus 4.6 — Maximum capability for complex reasoning, large architectural decisions, and demanding multi-file work

Why Sonnet 4.6

When Sonnet 4.6 was first introduced to Bolt in February 2026, Bolt positioned it as the most agentic Sonnet model Anthropic had ever released — one capable of planning multi-step features, parallelizing work, and executing complex tasks without requiring repeated manual intervention. Bolt noted at the time that Sonnet 4.6 "can hold a complex plan in its head and execute across multiple steps without losing the thread," handling multi-file refactors and coordinated front-and-backend changes that previously required switching to Opus.

Promoting Sonnet 4.6 to the default is a natural progression: what was once an opt-in upgrade is now the baseline experience for all Bolt users. The underlying model from Anthropic also benefits from a 1M-token context window (in beta), stronger long-context reasoning, and improvements to computer use — capabilities that make sustained, multi-session project work more reliable.

Impact on Existing Users

Users who had not manually selected a model will now build with Sonnet 4.6 by default, with no action required. Those who relied on Sonnet 4.5 or Opus 4.5 will need to adjust their workflows, as both models have been retired from the platform. Haiku 4.5 remains available for workflows where speed and token efficiency take priority over reasoning depth. For the most complex tasks, Opus 4.6 continues to be offered as the top-tier option.

Model selection remains limited to paid plans — free plan users do not have the ability to switch between models and will be served the platform default automatically.