ANTHROPIC · MODEL RELEASE TRACKER

Claude Sonnet 5

Claude Sonnet 5 is an Anthropic model. According to an AWS blog post, Claude Sonnet 5 holds FedRAMP Class D (formerly High) certification and DoD Impact Level 4 and 5 (IL4/IL5) authorization.

UPDATED 2026-10-06

CURRENT SNAPSHOT2/5 DIMENSIONS WITH DATA

The dimensions that change the decision.

PRICING

Not established from the available sources.

CONTEXT WINDOW

Not established from the available sources.

MODALITIES

Not established from the available sources.

BENCHMARKS

50-task Typed Domain Grounding benchmarkIn a fifty-task benchmark with Claude Sonnet 5, the Typed Domain Grounding approach achieved higher Structural Fidelity and a lower hallucination rate than two lenient external DSLs, although it had a lower first-try compile rate.2026-09-16

Benchmark on Typed Domain Grounding / DSL tasksIn a 50-task Typed Domain Grounding benchmark, Claude Sonnet 5 achieved higher structural fidelity and a lower hallucination rate than two lenient external DSL baselines, though it had a lower first-try compile rate in that evaluation.2026-09-16

AVAILABILITY

Certifications / authorizationsClaude Sonnet 5 holds FedRAMP Class D (formerly High) certification and DoD Impact Level 4 and 5 (IL4/IL5) authorization.2026-10-05

Availability on Amazon Bedrock (Seoul and Singapore, in-region inference)Claude Sonnet 5 is available on Amazon Bedrock with in-region inference in Seoul (ap-northeast-2) and in Singapore (ap-southeast-1) and is served on the bedrock-runtime endpoint.2026-09-30

Bedrock invocation: model ID and API supportOn Amazon Bedrock the model can be invoked with the direct model ID 'anthropic.claude-sonnet-5' and is compatible with Anthropic's Messages API as well as Amazon Bedrock's InvokeModel and Converse APIs (bedrock-runtime).2026-09-30

RELEASE TIMELINE

Published, source-backed release events only.

2026-09-24

1 SOURCES · IMPORTANCE 7.0

GitHub Security Lab releases LLM-driven Fuzzing Taskflow in seclab-taskflows-fuzzing repo

GitHub Security Lab published an autonomous, LLM-driven Fuzzing Taskflow built on its Taskflow Agent and available in the seclab-taskflows-fuzzing repository; the pipeline identifies entrypoints, generates harnesses, runs AFL++, measures coverage, triages crashes, and drafts vulnerability reports. The project uses MCP tools for execution, stores state in SQLite, and defaults to Claude Sonnet 5 as the model; it runs on a host (recommended in disposable environments) and is runnable via a Codespace.

→

VERIFIABLE FACTS

Every value stays attached to a source and date.

OTHER · Default model for GitHub Security Lab Fuzzing TaskflowDEVELOPER CLAIM

GitHub Security Lab states they 'use Claude Sonnet 5 by default' for their fuzzing Taskflow and reports it 'passed all of our internal tests without issues.'

BENCHMARKS · 50-task Typed Domain Grounding benchmarkDEVELOPER CLAIM

In a fifty-task benchmark with Claude Sonnet 5, the Typed Domain Grounding approach achieved higher Structural Fidelity and a lower hallucination rate than two lenient external DSLs, although it had a lower first-try compile rate.

AVAILABILITY · Certifications / authorizationsDEVELOPER CLAIM

Claude Sonnet 5 holds FedRAMP Class D (formerly High) certification and DoD Impact Level 4 and 5 (IL4/IL5) authorization.

AVAILABILITY · Availability on Amazon Bedrock (Seoul and Singapore, in-region inference)DEVELOPER CLAIM

Claude Sonnet 5 is available on Amazon Bedrock with in-region inference in Seoul (ap-northeast-2) and in Singapore (ap-southeast-1) and is served on the bedrock-runtime endpoint.

AVAILABILITY · Bedrock invocation: model ID and API supportDEVELOPER CLAIM

On Amazon Bedrock the model can be invoked with the direct model ID 'anthropic.claude-sonnet-5' and is compatible with Anthropic's Messages API as well as Amazon Bedrock's InvokeModel and Converse APIs (bedrock-runtime).

SAFETY · Use and internal safety testing by GitHub Security LabDEVELOPER CLAIM

GitHub Security Lab's Taskflow Agent uses Claude Sonnet 5 by default for their fuzzing taskflow, stating it 'passed all of our internal tests without issues', indicating it met their internal safety/guardrail checks for that pipeline.

BENCHMARKS · Benchmark on Typed Domain Grounding / DSL tasksDEVELOPER CLAIM

In a 50-task Typed Domain Grounding benchmark, Claude Sonnet 5 achieved higher structural fidelity and a lower hallucination rate than two lenient external DSL baselines, though it had a lower first-try compile rate in that evaluation.

WHAT CHANGED

Stored passport versions, without reconstructed history.

2026-10-06Passport updated7 facts

AVAILABILITY · Availability on Amazon Bedrock (India, cross-Region inference)− Claude Sonnet 5 is available on Amazon Bedrock in India via an India regional endpoint using geographic cross-Region inference (requests route between ap-south-1 and ap-south-2).2026-09-30

AVAILABILITY · Certifications / authorizations+ Claude Sonnet 5 holds FedRAMP Class D (formerly High) certification and DoD Impact Level 4 and 5 (IL4/IL5) authorization.2026-10-05

2026-10-03Passport updated7 facts

AVAILABILITY · Availability on Amazon Bedrock (India, cross-Region inference)+ Claude Sonnet 5 is available on Amazon Bedrock in India via an India regional endpoint using geographic cross-Region inference (requests route between ap-south-1 and ap-south-2).2026-09-30

AVAILABILITY · Bedrock invocation: model ID and API support+ On Amazon Bedrock the model can be invoked with the direct model ID 'anthropic.claude-sonnet-5' and is compatible with Anthropic's Messages API as well as Amazon Bedrock's InvokeModel and Converse APIs (bedrock-runtime).2026-09-30

AVAILABILITY · Availability on Amazon Bedrock (Seoul and Singapore, in-region inference)+ Claude Sonnet 5 is available on Amazon Bedrock with in-region inference in Seoul (ap-northeast-2) and in Singapore (ap-southeast-1) and is served on the bedrock-runtime endpoint.2026-09-30

SAFETY · Use and internal safety testing by GitHub Security Lab+ GitHub Security Lab's Taskflow Agent uses Claude Sonnet 5 by default for their fuzzing taskflow, stating it 'passed all of our internal tests without issues', indicating it met their internal safety/guardrail checks for that pipeline.2026-09-24

OTHER · Default model for GitHub Security Lab Fuzzing TaskflowGitHub Security Lab states they 'use Claude Sonnet 5 by default' for their fuzzing Taskflow and reports it 'passed all of our internal tests without issues.'→GitHub Security Lab states they 'use Claude Sonnet 5 by default' for their fuzzing Taskflow and reports it 'passed all of our internal tests without issues.'2026-09-24

BENCHMARKS · 50-task Typed Domain Grounding benchmarkIn a fifty-task benchmark with Claude Sonnet 5, the Typed Domain Grounding approach achieved higher Structural Fidelity and a lower hallucination rate than two lenient external DSLs, although it had a lower first-try compile rate.→In a fifty-task benchmark with Claude Sonnet 5, the Typed Domain Grounding approach achieved higher Structural Fidelity and a lower hallucination rate than two lenient external DSLs, although it had a lower first-try compile rate.2026-09-16

BENCHMARKS · Benchmark on Typed Domain Grounding / DSL tasks+ In a 50-task Typed Domain Grounding benchmark, Claude Sonnet 5 achieved higher structural fidelity and a lower hallucination rate than two lenient external DSL baselines, though it had a lower first-try compile rate in that evaluation.2026-09-16

2026-09-25Passport created2 facts

GitHub Security Lab releases LLM-driven Fuzzing Taskflow in seclab-taskflows-fuzzing repo

GitHub Security Lab states they 'use Claude Sonnet 5 by default' for their fuzzing Taskflow and reports it 'passed all of our internal tests without issues.'

In a fifty-task benchmark with Claude Sonnet 5, the Typed Domain Grounding approach achieved higher Structural Fidelity and a lower hallucination rate than two lenient external DSLs, although it had a lower first-try compile rate.

Claude Sonnet 5 holds FedRAMP Class D (formerly High) certification and DoD Impact Level 4 and 5 (IL4/IL5) authorization.

Claude Sonnet 5 is available on Amazon Bedrock with in-region inference in Seoul (ap-northeast-2) and in Singapore (ap-southeast-1) and is served on the bedrock-runtime endpoint.

On Amazon Bedrock the model can be invoked with the direct model ID 'anthropic.claude-sonnet-5' and is compatible with Anthropic's Messages API as well as Amazon Bedrock's InvokeModel and Converse APIs (bedrock-runtime).

GitHub Security Lab's Taskflow Agent uses Claude Sonnet 5 by default for their fuzzing taskflow, stating it 'passed all of our internal tests without issues', indicating it met their internal safety/guardrail checks for that pipeline.

In a 50-task Typed Domain Grounding benchmark, Claude Sonnet 5 achieved higher structural fidelity and a lower hallucination rate than two lenient external DSL baselines, though it had a lower first-try compile rate in that evaluation.

Optional analytics