Skip to main content

What’s New

See Anthropic’s model migration guide for a full overview of changes. Claude 5 Sonnet introduces three major changes versus Sonnet 4.6:
  1. Sampling parameters removed β€” temperature, top_p, and top_k are no longer supported and will be ignored
  2. Adaptive-only thinking β€” when reasoning is enabled the only supported mode is adaptive; thinking.budget_tokens is no longer supported and reasoning.max_tokens is ignored
  3. New 'xhigh' effort level β€” a new effort level between 'high' and 'max' via verbosity / output_config.effort
It also adds real-time cyber safeguards.

Sampling Parameters Removed

Claude 5 Sonnet no longer accepts temperature, top_p, or top_k. If you pass these parameters, they will be silently ignored β€” your request will still succeed, but the parameters will have no effect.

Adaptive-Only Thinking

Claude 5 Sonnet supports only adaptive thinking. On Sonnet 4.6, reasoning could be controlled via a token budget (reasoning.max_tokens / thinking.budget_tokens) or left adaptive; on Sonnet 5, budget-based thinking is removed and adaptive is the only remaining mode when reasoning is on. The new API default is adaptive thinking on at effort high. Reasoning is on by default on Sonnet 5: requests that don’t configure reasoning run with adaptive thinking enabled. Disable it explicitly with reasoning: { "enabled": false } if your integration assumes non-reasoning responses. Concretely on Sonnet 5:
  • reasoning.max_tokens is accepted but ignored
  • reasoning.effort maps to Anthropic’s output_config.effort (see Parameter Summary) rather than a thinking budget
  • thinking.budget_tokens is no longer supported upstream
To influence overall response effort (not reasoning-specific), use verbosity. It maps to Anthropic’s output_config.effort and applies whether or not reasoning is enabled.

New 'xhigh' Effort Level

A new 'xhigh' effort level is available between 'high' and 'max' via the verbosity parameter. This maps to Anthropic’s output_config.effort.
The full effort scale is now: low β†’ medium β†’ high β†’ xhigh β†’ max.
Sonnet 4.6 supported low, medium, high, and max. Sonnet 5 adds 'xhigh' between 'high' and 'max'. For models that don’t support a given level, it automatically falls back to the next supported level ('xhigh' β†’ 'high').

Cyber Safeguards

Like Opus 4.7 and 4.8, Claude 5 Sonnet ships with real-time cyber classifiers that block some high-risk dual-use activities. Affected requests are refused upstream by Anthropic; OpenRouter surfaces the refusal as-is. See Anthropic’s Real-time cyber safeguards on Claude for details.

Parameter Summary

With sampling parameters and reasoning budgets removed on Sonnet 5, output_config.effort is the remaining lever for influencing overall response effort. You can set it via verbosity, or via reasoning.effort when reasoning is enabled (verbosity wins if both are passed; 'minimal' maps to 'low', and 'none' disables reasoning entirely so no output_config.effort is sent).

Breaking Changes