Unofficial community list for Claude Sonnet 5. Not affiliated with Anthropic. All trademarks belong to their owners.
A curated list of resources for working with Claude Sonnet 5, the Anthropic model announced on June 30, 2026 and available on the Claude API as claude-sonnet-5. Anthropic calls it "the most agentic Sonnet model yet" and prices it at $2 per million input tokens and $10 per million output tokens. The links below come from the announcement, the platform docs and the CodeRabbit review that rank for the query; where a source names something without linking it, it appears in plain text.
Building a product that also needs image, video or audio generation? Try Synexa - one REST endpoint and Python SDK for FLUX, video and audio models, pay per run.
- Introducing Claude Sonnet 5 - The launch post: positioning against Sonnet 4.6 and Opus 4.8, pricing, plan availability.
- What's new in Claude Sonnet 5 - Model ID, the three behaviour changes from Sonnet 4.6, new tokenizer, tool support.
- Claude Sonnet 5 overview - The model page in the platform docs.
- Sonnet 5 migration guide - What breaks when moving from Sonnet 4.6.
- Claude Sonnet 5 System Card - The broader evaluation set the announcement points to.
- Models overview - Where the announcement sends developers for
claude-sonnet-5. - Pricing - Current rates; the announcement says the $2/$10 introductory pricing was made permanent.
- Try Claude - Sonnet 5 is the default model for Free and Pro plans and available to Max, Team and Enterprise, per the announcement.
- CLI, SDKs, and libraries - Client libraries for calling the API.
- Context windows - 1M tokens is both the default and the maximum on Sonnet 5; 128k max output.
- Thinking - Adaptive thinking is on by default; manual extended thinking now returns a 400.
- Effort - The setting the launch charts vary from medium up to xhigh.
- Model IDs and versioning - How the
claude-sonnet-5alias resolves.
- Claude Sonnet 5 review: Should you switch? - Hands-on review of how Sonnet 5 writes and reviews code; the verdict differs between the two.
- Choosing a model - Anthropic's guide to picking a tier.
- Optimizing for cost and intelligence - Background for reading the cost-performance curves in the launch post.
- Upgrade between model versions - The family-wide migration page.
- BrowseComp - The agentic search evaluation used in the launch charts.
- OSWorld-Verified - The computer use evaluation used in the launch charts.
- What is code review - Context for the review half of the CodeRabbit write-up.
- Browser use tool - Supported on Sonnet 5 on the Claude API and Google Cloud; not on Sonnet 4.6.
- Computer use tool - Sonnet 5 supports the stable
computer_toolset_20260801version;computer_20251124is still accepted. - Service tiers - Priority Tier is not available on Sonnet 5.
- Claude API skill - Keeps coding agents on current request shapes.
- Release notes - Platform changes since launch.
- Synexa - hosted model API for FLUX, video and audio, pay per run - Not a text model; the option when the missing piece next to Sonnet 5 is media generation behind one endpoint.
- Claude Opus 5 - The tier above in the docs navigation.
- Claude Fable 5.1 - The top model page in the current docs.
- Claude Haiku 4.5 - The small model in the line-up.
- Opus 4.8 - The model the launch post compares Sonnet 5 against; priced at $5/MTok input and $25/MTok output per the same post.
- Nemotron 3 Ultra - NVIDIA's open model, reviewed by CodeRabbit as "fast and to the point".
- Anthropic news - Where future announcements land.
- Model deprecations - Retirement schedule for older Sonnet versions.
- CodeRabbit blog - Reviews of Opus 4.8, Fable 5 and Nemotron 3 Ultra referenced in the Sonnet 5 write-up.
Open a pull request with a link, one line on why it belongs, and where it is referenced from.
Last reviewed: 2026-09-22