What are the verified coding and reasoning benchmarks for Claude Sonnet 5?

Claude Sonnet 5 achieves SWE-bench Verified: 75.8% (Thinking), LiveCodeBench: 67.2%, AIME: 87.5%. Verified sources: Claude Sonnet 5 Technical Report (Anthropic).

How does Claude Sonnet 5 compare to rival frontier coding models on SWE-bench and pricing?

Claude Sonnet 5 achieves SWE-bench Verified: 75.8% at $3/$15 per 1M tokens. Direct peer comparison: OpenAI o3 (71.7% SWE-bench, $5/$20); Gemini 3.1 Pro Preview (71% SWE-bench, $2.5/$10); xAI Grok 4.20 (58.5% SWE-bench, $2/$8); Claude Opus 4.6 (78% SWE-bench, $15/$75).

Does Swarm isolate Claude Sonnet 5 code edits from other agents?

Yes. Swarm provisions isolated Git worktrees for every session and subagent, preventing file race conditions and keeping your main working tree clean until verification passes.

Can Claude Sonnet 5 be assigned to specific subagent roles?

Yes. In ~/.swarm/config.json, you can configure Claude Sonnet 5 as Primary/Planner (DAG task decomposition), Coder (implementation in isolated worktrees), or Finder (read-only symbol search).

Claude Sonnet 5 — Specs, Context & Capabilities

Q: What is the recommended Swarm multi-agent topology for Claude Sonnet 5?

Swarm recommends Claude Sonnet 5 as Primary / Planner (Tier 1: Master Architect). The gold standard for primary orchestration. Manages full lifecycle from problem framing to task program DAG decomposition. Best paired with: Gemini 3.8 Flash for subagent swarms.

Q: How do I run Claude Sonnet 5 with Swarm?

Configure Claude Sonnet 5 via CLI using: swarm ctl model set --provider anthropic --model anthropic/claude-sonnet-5, or switch models interactively inside the terminal TUI with /models.

Claude Sonnet 5 Benchmarks & Agentic Accuracy

Verified primary research & developer preference metrics

Primary Sources Verified

SWE-bench Verified

75.8%

ThinkingAnthropic Preview ↗

LiveCodeBench

67.2%

ThinkingLiveCodeBench ↗

LMSYS Coding Elo

1435Elo

Coding ArenaLMSYS Arena ↗

AIME 2024 / 2025

87.5%

ThinkingAnthropic Preview ↗

Pricing ($ / 1M tokens)

$3/ $15

$0.4 cached readStandard API Pricing

What This Means in Real Software Development

75.8% autonomous pass rate on SWE-bench Verified real-world enterprise repositories.
Advanced multi-turn reasoning handles 50+ step development tasks with 0 context drift or goal abandonment.
Exceptional speed-to-intelligence ratio, generating production-grade patches with rapid execution.
Native tool-calling stability enables flawless multi-file patching across monorepos.

Swarm Multi-Agent Topology

Tier 1: Master ArchitectRole: Primary / Planner

Best Paired With: Gemini 3.8 Flash for subagent swarms

The gold standard for primary orchestration. Manages full lifecycle from problem framing to task program DAG decomposition.

Git Worktree Isolation Advantage:

Orchestrates up to 8 concurrent worktree subagents without race conditions.

What Developers & Practitioners Are Saying

Real-world field observations, production strengths, and developer consensus

Field Consensus

"Anthropic's premier workhorse coding engine, offering near-Opus reasoning depth at Sonnet execution speed."

Verified Real-World Strengths

75.8% on SWE-bench Verified
High-speed diff synthesis with zero extraneous explanation
Robust handling of large 200k context windows

Production Caveats & Gaps

Higher pricing than open-weights alternatives like DeepSeek V3

Primary Sources & Methodology Citations

Claude Sonnet 5 Technical Report — Anthropic (March 2026)

Head-to-Head Frontier Comparisons: Claude Sonnet 5 vs Rivals

Standardized, side-by-side benchmark & economics comparison against peer models

Flagship Comparative Matrix

Model	Role / Tier	SWE-bench	LiveCodeBench	LMSYS Elo	AIME	Input / Output
Claude Sonnet 5AnthropicCurrent	Tier 1: Master Architect	75.8%	67.2%	1435	87.5%	$3 / $15
OpenAI o3OpenAI Compare Claude Sonnet 5 vs OpenAI o3 →	Tier 1: Frontier Architect & Deductive Specialist	71.7%	76.2%	1415	92.4%	$5 / $20
Gemini 3.1 Pro PreviewGoogle Compare Claude Sonnet 5 vs Gemini 3.1 Pro Preview →	Tier 1: Frontier Large-Context Architect	71%	62.4%	1395	86.5%	$2.5 / $10
xAI Grok 4.20xAI Compare Claude Sonnet 5 vs xAI Grok 4.20 →	Tier 1: Frontier Systems Specialist	58.5%	54.2%	1375	76%	$2 / $8
Claude Opus 4.6Anthropic Compare Claude Sonnet 5 vs Claude Opus 4.6 →	Tier 1: Senior Principal Architect	78%	68.5%	1438	90%	$15 / $75

Explore All AI Model Comparisons Hub (Claude Sonnet 5 & Flagship Peers) →

Claude Sonnet 5 model overview

Claude Sonnet 5 is listed in the public Swarm catalog through Anthropic. This page summarizes the verified catalog fields used by Swarm without claiming real-time access for every provider account.

Provider: Anthropic
Family: Claude
Context window: 1M tokens
Thinking support: off · low · medium · high · xhigh
Availability status: Available
Catalog ID: anthropic/claude-sonnet-5

File input
Image input
Text input
Text output
Reasoning
Structured output
JSON
Batch

Run Claude Sonnet 5 in Swarm

Verified CLI workflow & multi-agent architecture

Quickstart Guide →

bash — swarm run

$swarm ctl model set --provider anthropic --model anthropic/claude-sonnet-5

# Or switch interactively inside the terminal TUI:

>/models

Subagent Role Matrix

Assign Claude Sonnet 5 to specialized system roles in ~/.swarm/config.json:

Primary / Planner: Orchestrates full task DAGs and multi-stage pipelines.
Coder Subagent: Generates code and unit tests inside private Git worktrees.
Finder Subagent: Performs read-only code search, symbol discovery, and mapping.

System Agent Roles →Why Swarm Architecture →