Claude Code v2.1.197: Claude Sonnet 5 Default, 1M Tokens, Promo Pricing
Claude Code v2.1.197 ships with Claude Sonnet 5 as the default model. The new model offers a native 1M-token context window and promotional pricing of $2 per million input tokens and $10 per million output tokens through August 31. Update to version 2.1.197 to get access.
Key Details
- Default Model Change: Claude Sonnet 5 is now the default in Claude Code. No manual model selection required after updating.
- Context Window: Native 1M-token context — same size as Claude 3.5 Sonnet, but now built in without chunking.
- Pricing Promotion: $2/Mtok input, $10/Mtok output until August 31. After the promo, pricing reverts to standard rates (not specified in source).
- Version Requirement: Must be on v2.1.197 to use Claude Sonnet 5 in Claude Code.
No other changes or fixes are documented in this release. The update is available now for all Claude Code users. To upgrade, run the standard update command for your setup.
📖 Read the full source: GitHub Claude-Code
👀 See Also
Claude Code v2.1.146: /code-review Command, Pagination Fix, Windows PowerShell Fix
Claude Code v2.1.146 renames /simplify to /code-review with optional effort level, fixes MCP pagination and Windows PowerShell tool, improves auto-updater reliability and diff rendering performance.
Claude-Code v2.1.108 adds prompt caching controls, recap feature, and slash command discovery
Claude-Code v2.1.108 introduces ENABLE_PROMPT_CACHING_1H and FORCE_PROMPT_CACHING_5M environment variables for cache TTL control, adds a session recap feature configurable via /config or /recap, and enables the model to discover built-in slash commands through the Skill tool.
Samsung Workers Demand Share of AI Chip Profits — What Developers Need to Know
Samsung's labor deal sets a precedent: 10.5% of operating profit from the semiconductor division goes to bonuses. A broader movement of workers across AI supply chains demanding a share of record profits.
User Reports Sonnet 4.6 Outperforms Opus 4.6 for Practical Coding Tasks
A developer testing Claude AI models found that Opus 4.6 produced over-engineered solutions with performance gaps, while Sonnet 4.6 delivered more careful, efficient fixes with lower token usage.