swarmagent.dev
GitHub
DocsModelsResourcesPricingNewsContactGitHub

Anthropic model

Claude Haiku 4.5

Claude Haiku 4.5 in the public Anthropic catalog.

Provider
Anthropic
Availability
Available
Model ID
claude-haiku-4-5-20251001
Snapshot
2026-10-07
Evaluation Verdict

Claude Haiku 4.5 delivers 52.4% on SWE-bench Verified with rapid execution, offering high-velocity, low-cost autonomous coding and AST navigation for Swarm subagent swarms.

Claude Haiku 4.5 Benchmarks & Agentic Accuracy

Verified primary research & developer preference metrics

Primary Sources Verified
SWE-bench Verified
52.4%
StandardAnthropic Technical Card ↗
LiveCodeBench
46.2%
StandardLiveCodeBench ↗
LMSYS Coding Elo
1320Elo
Coding ArenaLMSYS Arena ↗
AIME 2024 / 2025
58%
StandardAnthropic Technical Card ↗
Pricing ($ / 1M tokens)
$1/ $5
$0.1 cached readStandard API Pricing

What This Means in Real Software Development

  • Surpasses Claude 3.5 Sonnet's baseline coding score (52.4% vs 49.0%) while running at more than 2x the throughput.
  • Extremely economical $1.00/M input and $5.00/M output, slashing multi-agent parallel execution costs.
  • Sub-second response times (<350ms TTFT) make it the optimal interactive pairing engine.
  • Flawless file inspection and syntax patching inside git worktrees.

Swarm Multi-Agent Topology

Tier 3: High-Throughput Subagent EngineRole: Coder Subagent

Best Paired With: Claude 3.7 Sonnet as Lead Planner

The perfect balance of cost and coding competency for Stage 1 discovery and Stage 2 implementation jobs.

Git Worktree Isolation Advantage:

Dispatched across multiple worktrees concurrently to parallelize testing and linting.

What Developers & Practitioners Are Saying

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

Field Consensus

"Next-generation compact model from Anthropic, elevating speed and reasoning efficiency for autonomous workflows."

Verified Real-World Strengths

  • 52.4% on SWE-bench Verified at compact model pricing
  • Rapid generation speed with rapid response times
  • Superb adherence to strict type schemas and API interfaces

Production Caveats & Gaps

  • Higher per-token pricing than commodity open-weights models
Benchmark Reality Check

What the benchmark tests: Measures compact agentic coding, automated PR validation, and test generation.

Production caveat: Deploy as a high-speed validator subagent in multi-stage Task Programs.

Primary Sources & Methodology Citations

Claude Haiku 4.5 model overview

Claude Haiku 4.5 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
200K tokens
Thinking support
off
Availability status
Available
Catalog ID
anthropic/claude-haiku-4-5-20251001
  • File input
  • Image input
  • Text input
  • Text output
  • Reasoning
  • Structured output
  • JSON
  • Batch

Run Claude Haiku 4.5 in Swarm

Verified CLI workflow & multi-agent architecture

Quickstart Guide →
bash — swarm run
$swarm ctl model set --provider anthropic --model anthropic/claude-haiku-4-5-20251001

# Or switch interactively inside the terminal TUI:

>/models

Subagent Role Matrix

Assign Claude Haiku 4.5 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 →
← All Anthropic models

Public catalog snapshot. Catalog availability is not a real-time provider status or an access guarantee.