Sonnet 5
The latest version in Anthropic's Claude Sonnet series, positioned among leading large language models. It demonstrates strong performance across multiple task types, including code generation and multi-turn dialogue, representing Anthropic's mid-tier model balancing capability and efficiency.
Core Facts
Sonnet 5.5 runs more than 30% faster than Sonnet 5.
The company states that Sonnet 5 shows significant improvements over Sonnet 4.6 in reasoning, tool use, coding, and knowledge work, with overall performance approaching OPUS 4.8 but at a lower price.
On the SWE Bench Pro programming benchmark, Sonnet 5 achieved 63.2% and Opus 4.8 achieved 69.2%.
Timeline (last 90 days)
In the final build stage, the author used Sonnet 5, and it was completed with just a single prompt.
Sonnet 5's per-task cost is about eight times that of Claude Haiku 5.5.
Because Sonnet 5 burns through tokens aggressively, its actual total cost to complete tasks is even higher than that of Astra.
The developer employs a sub-agent workflow, using Opus 5.5 for the main game-building tasks and delegating trivial commit and screenshot tasks to the lighter-weight Sonnet 5.
6.1 Sol reads an average of approximately 110,000 tokens per request, whereas Sonnet 5 requires approximately 360,000 tokens.
Anthropic officially positions Sonnet 5.5 as an upgrade to Sonnet 5, with a 30% speed improvement and cost reductions of up to 30% for most workloads.
GPT-4.1 Sol reads an average of approximately 110,000 tokens per request, while Sonnet 5 requires 360,000 tokens.
The earlier Sonnet 5 had mediocre benchmark scores and suffered from runaway cost increases on complex tasks.
Sonnet 5.5 is the first Sonnet model released with cybersecurity safeguards, where high-risk cybersecurity tasks will explicitly fall back to Sonnet 5
On Terminal Bench, mid-tier Sonnet 5.5 scores 28.8% at approximately $0.83 per task, while Sonnet 5's highest tier only reaches 10.3% at approximately $11.6 per task
24 more timeline events
All Facts (20)
Sonnet 5.5 runs more than 30% faster than Sonnet 5.
80%VerifiedThe company states that Sonnet 5 shows significant improvements over Sonnet 4.6 in reasoning, tool use, coding, and knowledge work, with overall performance approaching OPUS 4.8 but at a lower price.
70%VerifiedOn the SWE Bench Pro programming benchmark, Sonnet 5 achieved 63.2% and Opus 4.8 achieved 69.2%.
70%Partially VerifiedSonnet 5.5 is the first Sonnet model released with cybersecurity safeguards, where high-risk cybersecurity tasks will explicitly fall back to Sonnet 5
65%UnverifiedOn Terminal Bench, mid-tier Sonnet 5.5 scores 28.8% at approximately $0.83 per task, while Sonnet 5's highest tier only reaches 10.3% at approximately $11.6 per task
60%UnverifiedIn the final build stage, the author used Sonnet 5, and it was completed with just a single prompt.
50%UnverifiedSonnet 5's per-task cost is about eight times that of Claude Haiku 5.5.
50%UnverifiedBecause Sonnet 5 burns through tokens aggressively, its actual total cost to complete tasks is even higher than that of Astra.
50%UnverifiedThe developer employs a sub-agent workflow, using Opus 5.5 for the main game-building tasks and delegating trivial commit and screenshot tasks to the lighter-weight Sonnet 5.
50%Unverified6.1 Sol reads an average of approximately 110,000 tokens per request, whereas Sonnet 5 requires approximately 360,000 tokens.
50%UnverifiedAnthropic officially positions Sonnet 5.5 as an upgrade to Sonnet 5, with a 30% speed improvement and cost reductions of up to 30% for most workloads.
50%UnverifiedGPT-4.1 Sol reads an average of approximately 110,000 tokens per request, while Sonnet 5 requires 360,000 tokens.
50%UnverifiedThe earlier Sonnet 5 had mediocre benchmark scores and suffered from runaway cost increases on complex tasks.
50%UnverifiedSessions behind a custom `ANTHROPIC_BASE_URL` will now use models supporting a 1-million-token context window (Opus 4.7+, Sonnet 5+, Fable); if the gateway truncates at 200K, `/autocompact 200k` can be run as a workaround.
50%UnverifiedClaude Sonnet 5.5 has improved efficiency on coding tasks and agentic use cases compared to its predecessor Sonnet 5.
50%UnverifiedOn the FrontierCode 1.1 benchmark, Sonnet 5.5 achieved notable improvement compared to its predecessor Sonnet 5.
50%UnverifiedAccording to Anthropic's automated behavioral audit, Sonnet 5.5 outperforms Sonnet 5 on most alignment and honesty metrics
50%UnverifiedIn a Wikipedia Speedrun demonstration, Jev completed five hops in just half a second, while Tara 5.6, Haiku 4.5, and Sonnet 5 took approximately 4 seconds, 5 seconds, and 5 seconds respectively.
50%UnverifiedIn benchmark tests, official demonstrations show Jev on par with Luna, Tara, and Sonnet 5—and even surpassing Opus 5—on metrics such as type safety, but at a fraction of the price.
50%UnverifiedIn the blogger's Claire weighted index (70% personal taste + 30% LLM judgment), Grok 4.6 surpassed Sonnet 5 and Opus 5, ranking just behind GPT-5.6.
50%