What are the verified coding and reasoning benchmarks for GPT-5.5?

GPT-5.5 achieves SWE-bench Verified: 78.5% (Thinking), LiveCodeBench: 82%, AIME: 96.5%. Verified sources: OpenAI GPT-5.5 Milestone Announcement (OpenAI).

How do I run GPT-5.5 with Swarm?

Configure GPT-5.5 via CLI using: swarm ctl model set --provider codex --model codex/gpt-5.5, or switch models interactively inside the terminal TUI with /models.

Does Swarm isolate GPT-5.5 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 GPT-5.5 be assigned to specific subagent roles?

Yes. In ~/.swarm/config.json, you can configure GPT-5.5 as Primary/Planner (DAG task decomposition), Coder (implementation in isolated worktrees), or Finder (read-only symbol search).

GPT-5.5 — Specs, Context & Capabilities

Q: What is the recommended Swarm multi-agent topology for GPT-5.5?

Swarm recommends GPT-5.5 as Primary / Planner (Tier 1: Master Autonomous Architect). The premier orchestrator for enterprise-scale software engineering tasks. Best paired with: Gemini 3.8 Flash for subagent implementation.

GPT-5.5 Benchmarks & Agentic Accuracy

Verified primary research & developer preference metrics

Primary Sources Verified

SWE-bench Verified

78.5%

ThinkingOpenAI Preview ↗

LiveCodeBench

82%

ThinkingLiveCodeBench ↗

LMSYS Coding Elo

1455Elo

Coding ArenaLMSYS Arena ↗

AIME 2024 / 2025

96.5%

ThinkingOpenAI Preview ↗

Pricing ($ / 1M tokens)

$3/ $12

$0.7 cached readStandard API Pricing

What This Means in Real Software Development

78.5% on SWE-bench Verified sets a new high-water mark for autonomous repository engineering.
Multi-modal vision and code synthesis enables end-to-end web application refactoring from UI mocks.
Operates with high reasoning depth and zero context degradation.
Unparalleled accuracy in Task Program DAG decomposition and integration barrier validation.

Swarm Multi-Agent Topology

Tier 1: Master Autonomous ArchitectRole: Primary / Planner

Best Paired With: Gemini 3.8 Flash for subagent implementation

The premier orchestrator for enterprise-scale software engineering tasks.

Git Worktree Isolation Advantage:

Supervises multi-agent worktrees and guarantees clean branch integration.

What Developers & Practitioners Are Saying

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

Field Consensus

"Refined iteration of the GPT-5 family, optimizing token efficiency and boosting complex multi-step reasoning accuracy."

Verified Real-World Strengths

78.5% on SWE-bench Verified
High throughput with near-frontier reasoning quality
Exceptional JSON schema and type contract fidelity

Production Caveats & Gaps

Output tokens can scale quickly on open-ended refactoring prompts
Still benefits from worktree isolation to catch edge-case regressions

Primary Sources & Methodology Citations

OpenAI GPT-5.5 Milestone Announcement — OpenAI (April 2026)

GPT-5.5 model overview

GPT-5.5 is listed in the public Swarm catalog through OpenAI Codex (OAuth). This page summarizes the verified catalog fields used by Swarm without claiming real-time access for every provider account.

Provider: OpenAI Codex (OAuth)
Family: Gpt 5
Context window: 272K tokens
Thinking support: off · low · medium · high · xhigh
Availability status: Available
Catalog ID: codex/gpt-5.5

Image input
Text input
Text output
Reasoning
Tools

Run GPT-5.5 in Swarm

Verified CLI workflow & multi-agent architecture

Quickstart Guide →

bash — swarm run

$swarm ctl model set --provider codex --model codex/gpt-5.5

# Or switch interactively inside the terminal TUI:

>/models

Subagent Role Matrix

Assign GPT-5.5 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 →