Anthropic: Claude Sonnet 4.5

Claude Sonnet 4.5 represents a significant evolution in the model family, specifically engineered to excel in coding workflows and real-world agentic tasks. It is designed to handle complex system design, code security, and strict specification adherence, achieving high performance on benchmarks like SWE-bench Verified. By focusing on extended autonomous operation, the model maintains task continuity across sessions, making it a reliable choice for long-running processes that require consistent, fact-based progress tracking.

The model introduces refined capabilities in tool orchestration and memory management, allowing for speculative parallel execution and more efficient context handling. These advancements enable the model to maintain awareness of token usage across multiple tool calls, which is essential for multi-context environments. With its enhanced domain knowledge and improved reliability, the model is well-suited for demanding professional applications in software engineering, cybersecurity, and financial analysis, providing a robust foundation for developers building scalable and intelligent AI products.

Heliconeclaude-4.5-sonnetclaude-sonnet

Open provider page Provider docs

Quick Info

Provider: Helicone
Model key: claude-4.5-sonnet
Release date: Sep 29, 2025
Last updated: Sep 29, 2025
Knowledge cutoff: 2025-09
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $3.00
Output token cost: $15.00

Limits

Output tokens: 64,000 tokens
Context window: 200,000 tokens

Transparent token rates

Compare Anthropic: Claude Sonnet 4.5 pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about Anthropic: Claude Sonnet 4.5

Helicone

CoverageComparison

Rumors point to Sonnet 5 at half the cost of Opus 4.5 and stronger code tools, guiding smarter picks for teams watching pricing.

Helicone

CoverageBenchmark

Claude Sonnet 4.6 offers a 1-million-token context window in beta for codebases or contracts, helping you review large files faster.

Helicone

Coverage

Sonnet 4.6 adds adaptive thinking and browser task gains with 4x higher token use than Sonnet 4.5, budget planning changes by task type.

Helicone

CoverageRelease Notes

Discover more about what's new at AWS with Amazon Bedrock increases default quotas for Anthropic’s Claude Sonnet 4.5 model in AWS GovCloud (US)

Helicone

Coverage

Epoch AI's profile confirms Claude Sonnet 4.5 as a closed-weights Anthropic model released September 29, 2025, assigning it an Epoch Capabilities Index score of 147 and a ranking of 64th out of 251 tracked models, having debuted at 4th place among 166 models at launch. The benchmark scorecard compares it against Claude Opus 4.1, GPT-5, and Grok 4 across mathematics, science, software engineering, and world-knowledge evaluations. Provider-listed pricing stands at $3 per million input tokens and $15 per million output tokens. Headline results include 83% on SWE-bench Verified, 96% on GPQA Diamond, 100% on OTIS Mock AIME 2024-2025, 98% on FrontierMath Tier 4, and 76% on SimpleQA Verified, with the strongest agentic and reasoning scores among the compared alternatives. Epoch notes the knowledge cutoff and training-compute details as unknown, reflecting the closed-weights nature of the release. The page serves as an independent technical reference for evaluating Claude Sonnet 4.5's standing relative to peer frontier models.

Helicone

CoverageRelease Notes

GitHub Copilot began a gradual rollout of Claude Sonnet 4.5 on September 29, 2025, making Anthropic's newest coding-focused model available in the Copilot Chat model picker for Pro, Pro+, Business, and Enterprise users. The rollout spans Visual Studio Code (chat, ask, edit, and agent modes), github.com, GitHub Mobile on iOS and Android, and the Copilot CLI starting with version 0.0.329, where users can switch models via the /model command. Early GitHub testing highlighted major upgrades in tool orchestration, context editing, and domain-specific capabilities for real-world agents. Enterprise and Business administrators must enable the Claude Sonnet 4.5 policy in Copilot settings before users can select it, while Free, Pro, and Pro+ users in VS Code can bring their own Anthropic API key via the Manage Models option in the picker. GitHub positioned the model as Anthropic's most advanced offering for coding and agentic workloads at the time of the public preview announcement. Related Copilot changelog entries from the same week referenced additional model additions including GPT-6 variants and Grok 4.7.