Back to News
ProductAI Understanding briefing

Anthropic launches Claude Sonnet 5.5, a faster, lower‑cost AI model

Anthropic unveiled Claude Sonnet 5.5, a new model that runs 30 percent faster than its predecessor and cuts per‑task costs by up to 30 percent, while delivering performance close to the flagship Opus 5.5.

4 min readRead the linked source
Source-provided image accompanying Anthropic launches Claude Sonnet 5.5, a faster, lower‑cost AI model
Source referenceSource recorded
Publisher
suaragarut.id
Source link
suaragarut.idhttps://suaragarut.id/en/anthropic-launches-claude-sonnet-5-5
Source type
Linked source — primary-source status has not been established.
Also cited

Story last revised

ContextUnderstand this in 60 seconds

Start here

Key terms

API (Application Programming Interface)
A structured way for one software system to send requests to and receive responses from another system.
Benchmark
A standardized test or dataset used to measure and compare model performance.
Latency
The time between sending a request and receiving the model's output.
Test yourselfWhat is AI? Quiz

What changed since publication

  1. First published
  2. The article adds concrete benchmark scores (Terminal‑Bench 4.0, GDPval‑AA, FrontierCode), confirms unchanged API pricing but lower effective costs, announces upcoming Claude Haiku 5.5, and notes new cybersecurity safeguards, extending the details of the Claude Sonnet 5.5 launch.

What happened

Anthropic announced the release of Claude Sonnet 5.5, the second model in its Claude 5.5 family. The new model is advertised as more than 30 percent faster than Claude Sonnet 5 and claims to reduce overall task expenses by up to 30 percent. testing reported notable gains: on the Terminal‑Bench 4.0 coding evaluation the model scored 70.6 percent (up from 10.3 percent for Sonnet 5); on the GDPval‑AA workplace‑task benchmark it fell within two points of Opus 5.5; and on FrontierCode at high‑effort settings it added ten points over Sonnet 5. The model also demonstrated new capabilities such as navigating Pokémon Red using only screenshots. Pricing for the API remains unchanged at $2 per million input tokens, $10 per million output tokens, and $0.20 per million cache‑read tokens, but lower token usage translates to lower per‑task costs. Anthropic added cybersecurity safeguards comparable to those in Opus 5, marking the first Sonnet variant with specialized high‑risk request protections. A high‑volume Claude Haiku 5.5 is slated for release in the coming weeks.

Anthropic released Claude Sonnet 5.5 as the second model in its Claude 5.5 series, emphasizing speed and cost efficiency for everyday AI tasks. The company states the model runs over 30 percent faster than the prior Sonnet 5 while delivering comparable or better results on a range of benchmarks.

data supplied by Anthropic shows the model achieving a 70.6 percent score on Terminal‑Bench 4.0, a coding‑agent evaluation, a dramatic increase from Sonnet 5’s 10.3 percent. On the GDPval‑AA workplace‑task benchmark, Sonnet 5.5 trails Opus 5.5 by only two points, indicating near‑top‑tier performance on real‑world occupational tasks.

In programming tests, Sonnet 5.5 outperformed its predecessor by ten points on FrontierCode under high‑effort settings and reduced task costs by roughly fifteen‑fold. The model also became the first Sonnet variant to successfully navigate Pokémon Red using only screenshot inputs, showcasing improved long‑horizon reasoning.

Pricing for the API remains the same as for Sonnet 5 ($2 M input tokens, $10 M output tokens, $0.20 M cache reads), but the reduced token consumption per task translates into lower effective costs. Anthropic also introduced cybersecurity safeguards similar to those in Opus 5, marking the first Sonnet model with specialized high‑risk request protections.

Source details: suaragarut.id ↗

Why it matters

The launch of Claude Sonnet 5.5 matters because it offers enterprises a faster, cheaper alternative to higher‑priced models while approaching the performance of Anthropic’s flagship Opus 5.5. The speed and cost improvements can lower operating expenses for developers and businesses that run large‑scale workloads, such as code generation, document creation, and long‑horizon knowledge work. results suggest the model narrows the gap between mid‑tier and top‑tier offerings, potentially reshaping pricing dynamics in the competitive generative‑AI market. The inclusion of dedicated cybersecurity safeguards also signals Anthropic’s focus on responsible deployment, addressing growing concerns about misuse of powerful language models.

The performance‑cost balance of Claude Sonnet 5.5 gives enterprises a more economical option for large‑scale AI workloads, potentially lowering total cost of ownership for applications such as code assistance, document generation, and complex data‑flow analysis.

By narrowing the performance gap with Opus 5.5, Sonnet 5.5 may shift market expectations for mid‑tier models, prompting competitors to reassess pricing and capability trade‑offs.

The addition of dedicated cybersecurity safeguards reflects Anthropic’s ongoing effort to mitigate misuse risks, a factor increasingly scrutinized by regulators and corporate risk officers.

The announced upcoming release of Claude Haiku 5.5 suggests Anthropic is building a tiered portfolio that can serve diverse volume and requirements, which could influence enterprise AI strategy and vendor selection.

Interactive Mechanism

Interactive Mechanism: How It Actually Works

Explore the underlying technology behind this development interactively.

Thinking Budget (Test-Time Tokens):1,024 tokens
Complex Accuracy79%Math & Code Logic
Latency3.2sTime to first full output
Inference Cost$0.0092Per query estimated
Reasoning StyleStep VerificationInternal chain depth
Active Thinking Trace:
1Deconstruct user problem into formal constraints
2Propose candidate hypotheses & step-by-step calculation
3Self-correction: Backtrack and refute subtle edge cases
4Exhaustive consistency check & final output synthesis
Core takeaway: Test-time compute fundamentally changes AI economics. Instead of only scaling during pre-training, giving reasoning models more tokens at inference time allows them to systematically solve PhD-level STEM problems.
Interactive Concept Check+10 Points
What is AI? Quiz

Which description best fits "narrow AI", the kind of AI in use today?

What to watch next

Future developments to monitor include the upcoming Claude Haiku 5.5, which is positioned as a high‑volume, lower‑cost model for massive token‑throughput scenarios. Observers should watch for any changes to API pricing or token‑usage discounts that could further affect cost competitiveness. Adoption metrics from early corporate testers—especially in software development and gaming pipelines—will indicate whether the performance claims translate into real‑world efficiency gains. Finally, the impact of Anthropic’s new safety safeguards on user experience and incident rates will be a key indicator of the model’s suitability for enterprise‑critical applications.

Release and performance of Claude Haiku 5.5, especially its cost‑per‑token metrics and throughput capabilities.

Any adjustments to API pricing or token‑usage discounts that could further differentiate Sonnet 5.5 from competing models.

Adoption rates and feedback from early corporate users, particularly in software development and gaming sectors, to gauge real‑world efficiency gains.

Effectiveness of the newly added cybersecurity safeguards in preventing high‑risk request abuse and their impact on user experience.

Related guides & quizzes

What is AI?AI Models ExplainedFuture of AIAI EthicsTest what you know — try a free AI quizLook up an AI term in our glossaryFollow the AI model release tracker

Updates and corrections

This canonical story is updated in place when the developing event materially changes. Its URL and original publication date never change.

  • The article adds concrete benchmark scores (Terminal‑Bench 4.0, GDPval‑AA, FrontierCode), confirms unchanged API pricing but lower effective costs, announces upcoming Claude Haiku 5.5, and notes new cybersecurity safeguards, extending the details of the Claude Sonnet 5.5 launch.
See the public corrections log
Found this useful?