Per-task costSonnet 5.5 reduces per-task costs by up to about 30% compared with Sonnet 5 through more efficient token usage and faster execution.
ANTHROPIC · MODEL RELEASE TRACKER
Claude Sonnet 5.5
Claude Sonnet 5.5 is a Claude-series model from Anthropic. Anthropic lists it as a generally available, 'most capable' model; it is included in the expanded Cyber Verification Program (CVP) access tiers and — in its generally available form — has conservative cyber safeguards that block most cyber work.CURRENT SNAPSHOT3/5 DIMENSIONS WITH DATA
The dimensions that change the decision.
Not established from the available sources.
Not established from the available sources.
Benchmark performance (selected results)Reported benchmark results include: Terminal-Bench 4.0 — Sonnet 5.5: 70.6% vs Sonnet 5: 10.3%; CursorBench 4.0 — Sonnet 5.5: 55.5% vs Sonnet 5: 34.1% (Sonnet 5.5 two points below Opus 5.5 at 57.8%); FrontierCode 1.1 (High) — Sonnet 5.5 scores ~10 points above Sonnet 5 at roughly one-fifteenth the cost per task (per Anthropic reporting).
AvailabilityListed by Anthropic as a generally available model and included among models made accessible through the expanded Cyber Verification Program (CVP) access tiers for qualifying security professionals.
Availability on AWS GovCloud (US) via Amazon BedrockClaude Sonnet 5.5 is available in AWS GovCloud (US) Regions on Amazon Bedrock.
RELEASE TIMELINE
Published, source-backed release events only.
VERIFIABLE FACTS
Every value stays attached to a source and date.
AVAILABILITY · AvailabilityDEVELOPER CLAIM
Listed by Anthropic as a generally available model and included among models made accessible through the expanded Cyber Verification Program (CVP) access tiers for qualifying security professionals.
CAPABILITIES · Expected performance and improvementsDEVELOPER CLAIM
Reported to be part of the 5.5 series and expected to deliver similar performance, efficiency, and safety improvements to Opus 5.5 (Anthropic said Sonnet 5.5 will have similar gains). This is reported as an expectation, not a released capability.
RELEASE · Release dateDEVELOPER CLAIM
Claude Sonnet 5.5 was announced and released on 2026-09-28.
AVAILABILITY · Availability on AWS GovCloud (US) via Amazon BedrockDEVELOPER CLAIM
Claude Sonnet 5.5 is available in AWS GovCloud (US) Regions on Amazon Bedrock.
CAPABILITIES · Speed and token efficiencyDEVELOPER CLAIM
Anthropic and partners report Sonnet 5.5 generates output more than 30% faster than Sonnet 5 and uses tokens more efficiently, reducing token burn relative to its predecessor.
PRICING · Per-task costDEVELOPER CLAIM
Sonnet 5.5 reduces per-task costs by up to about 30% compared with Sonnet 5 through more efficient token usage and faster execution.
BENCHMARKS · Benchmark performance (selected results)DEVELOPER CLAIM
Reported benchmark results include: Terminal-Bench 4.0 — Sonnet 5.5: 70.6% vs Sonnet 5: 10.3%; CursorBench 4.0 — Sonnet 5.5: 55.5% vs Sonnet 5: 34.1% (Sonnet 5.5 two points below Opus 5.5 at 57.8%); FrontierCode 1.1 (High) — Sonnet 5.5 scores ~10 points above Sonnet 5 at roughly one-fifteenth the cost per task (per Anthropic reporting).
CAPABILITIES · Target use casesDEVELOPER CLAIM
Sonnet 5.5 is positioned for well-scoped everyday work such as focused coding (including agentic coding), short document and slide creation, spreadsheet edits, and routine knowledge-work tasks where fast execution and lower cost per task are priorities.
SAFETY · Cyber safeguards / limitationDEVELOPER CLAIM
Anthropic states that the generally available Claude Sonnet 5.5 has conservative cyber safeguards that block most cyber work.
LIMITATIONS · Tier and relative powerDEVELOPER CLAIM
Anthropic describes Sonnet as a mid-tier model that is less powerful than Opus (which is aimed at higher-judgment tasks) but in many cases more agile and cost-effective for agentic and high-throughput everyday work.
CAPABILITIES · Supports AI-assisted development for regulated/compliance workloadsDEVELOPER CLAIM
The availability of Claude Sonnet 5.5 on Bedrock in AWS GovCloud (US) is described as an on-ramp for AI-assisted development for workloads with regulatory or compliance requirements, including ITAR.
CAPABILITIES · Capability / program inclusionDEVELOPER CLAIM
Named by Anthropic as one of its 'most capable' models and listed among models (with Claude Opus 5.5 and Claude Mythos 5.1) that are available in CVP tiers offering reduced blocking classifiers to vetted security teams.
WHAT CHANGED
Stored passport versions, without reconstructed history.
AVAILABILITY · AvailabilityAnnounced as an upcoming model in the 'Claude 5.5' family and said to be released "in the coming weeks" (i.e., expected imminent release).→Listed by Anthropic as a generally available model and included among models made accessible through the expanded Cyber Verification Program (CVP) access tiers for qualifying security professionals.
SAFETY · Cyber safeguards / limitation+ Anthropic states that the generally available Claude Sonnet 5.5 has conservative cyber safeguards that block most cyber work.
SAFETY · FedRAMP Class D certification on Amazon Bedrock− Claude Sonnet 5.5 holds FedRAMP Class D certification on Amazon Bedrock (report notes to verify current model certification status).
CAPABILITIES · Capability / program inclusion+ Named by Anthropic as one of its 'most capable' models and listed among models (with Claude Opus 5.5 and Claude Mythos 5.1) that are available in CVP tiers offering reduced blocking classifiers to vetted security teams.
AVAILABILITY · Availability on AWS GovCloud (US) via Amazon Bedrock+ Claude Sonnet 5.5 is available in AWS GovCloud (US) Regions on Amazon Bedrock.
AVAILABILITY · Availability on clouds and platforms− Sonnet 5.5 is available on Amazon Bedrock / Claude Platform on AWS and is also offered on Google Cloud and Microsoft Azure. On AWS it integrates with regional data residency and AWS controls such as IAM, CloudTrail and CloudWatch, and usage appears on the AWS bill.
SAFETY · Cybersecurity and distillation safeguards− Anthropic reports that Sonnet 5.5 includes new safeguards against cybersecurity risks and distillation attacks and that Sonnet will be subject to the same cyber safeguards as other higher-tier models (e.g., Fable and Opus).
SAFETY · FedRAMP Class D certification on Amazon Bedrock+ Claude Sonnet 5.5 holds FedRAMP Class D certification on Amazon Bedrock (report notes to verify current model certification status).
CAPABILITIES · Supports AI-assisted development for regulated/compliance workloads+ The availability of Claude Sonnet 5.5 on Bedrock in AWS GovCloud (US) is described as an on-ramp for AI-assisted development for workloads with regulatory or compliance requirements, including ITAR.
BENCHMARKS · Benchmark performance (selected results)+ Reported benchmark results include: Terminal-Bench 4.0 — Sonnet 5.5: 70.6% vs Sonnet 5: 10.3%; CursorBench 4.0 — Sonnet 5.5: 55.5% vs Sonnet 5: 34.1% (Sonnet 5.5 two points below Opus 5.5 at 57.8%); FrontierCode 1.1 (High) — Sonnet 5.5 scores ~10 points above Sonnet 5 at roughly one-fifteenth the cost per task (per Anthropic reporting).
AVAILABILITY · Availability on clouds and platforms+ Sonnet 5.5 is available on Amazon Bedrock / Claude Platform on AWS and is also offered on Google Cloud and Microsoft Azure. On AWS it integrates with regional data residency and AWS controls such as IAM, CloudTrail and CloudWatch, and usage appears on the AWS bill.
PRICING · Per-task cost+ Sonnet 5.5 reduces per-task costs by up to about 30% compared with Sonnet 5 through more efficient token usage and faster execution.
SAFETY · Cybersecurity and distillation safeguards+ Anthropic reports that Sonnet 5.5 includes new safeguards against cybersecurity risks and distillation attacks and that Sonnet will be subject to the same cyber safeguards as other higher-tier models (e.g., Fable and Opus).
RELEASE · Release date+ Claude Sonnet 5.5 was announced and released on 2026-09-28.
CAPABILITIES · Speed and token efficiency+ Anthropic and partners report Sonnet 5.5 generates output more than 30% faster than Sonnet 5 and uses tokens more efficiently, reducing token burn relative to its predecessor.
CAPABILITIES · Target use cases+ Sonnet 5.5 is positioned for well-scoped everyday work such as focused coding (including agentic coding), short document and slide creation, spreadsheet edits, and routine knowledge-work tasks where fast execution and lower cost per task are priorities.
LIMITATIONS · Tier and relative power+ Anthropic describes Sonnet as a mid-tier model that is less powerful than Opus (which is aimed at higher-judgment tasks) but in many cases more agile and cost-effective for agentic and high-throughput everyday work.