What are the verified coding and reasoning benchmarks for GPT 5?

GPT 5 achieves SWE-bench Verified: 69.5% (Frontier Thinking), LiveCodeBench: 74%, AIME: 91.2%. Verified sources: OpenAI GPT-5 Technical Architecture (OpenAI).

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

GPT 5 achieves SWE-bench Verified: 69.5% at $2.5/$10 per 1M tokens. Direct peer comparison: Claude 3.7 Sonnet (70.3% SWE-bench, $3/$15); Gemini 3.1 Pro Preview (71% SWE-bench, $2.5/$10); xAI Grok 4.20 (58.5% SWE-bench, $2/$8); OpenAI o3 (71.7% SWE-bench, $5/$20).

How do I run GPT 5 with Swarm?

Configure GPT 5 via CLI using: swarm ctl model set --provider openai --model openai/gpt-5, or switch models interactively inside the terminal TUI with /models.

Does Swarm isolate GPT 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 be assigned to specific subagent roles?

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

GPT 5 — Specs, Context & Capabilities

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

Swarm recommends GPT 5 as Primary / Planner (Tier 1: Master Autonomous Architect). High-confidence planning eliminates hallucinated stages and coordinates complex multi-worker Task Programs effortlessly. Best paired with: Gemini 3.8 Flash for subagent swarms.

GPT 5 Benchmarks & Agentic Accuracy

Verified primary research & developer preference metrics

Primary Sources Verified

SWE-bench Verified

69.5%

Frontier ThinkingOpenAI Technical Report ↗

LiveCodeBench

74%

ThinkingLiveCodeBench ↗

LMSYS Coding Elo

1418Elo

Coding ArenaLMSYS Arena ↗

AIME 2024 / 2025

91.2%

ThinkingOpenAI Technical Report ↗

Pricing ($ / 1M tokens)

$2.5/ $10

$0.6 cached readStandard API Pricing

What This Means in Real Software Development

Frontier autonomous coding score: 69.5% on SWE-bench Verified across complex GitHub repositories.
91.2% on AIME 2024 and 84.0% on GPQA Diamond prove supreme multi-step scientific and mathematical reasoning.
Dynamic thinking mode dynamically allocates compute based on problem complexity, optimizing cost and latency.
Full monorepo AST awareness allows refactoring cross-package dependencies seamlessly.

Swarm Multi-Agent Topology

Tier 1: Master Autonomous ArchitectRole: Primary / Planner

Best Paired With: Gemini 3.8 Flash for subagent swarms

High-confidence planning eliminates hallucinated stages and coordinates complex multi-worker Task Programs effortlessly.

Git Worktree Isolation Advantage:

Supervises parallel Git worktrees and verifies build assertions before promoting code to dev.

What Developers & Practitioners Are Saying

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

Field Consensus

"OpenAI's major generational platform update. Exceptionally balanced between raw reasoning depth, fast throughput, and reliable tool calling."

Verified Real-World Strengths

69.5% on SWE-bench Verified with zero configuration overhead
Crisp, concise code explanations without unnecessary boilerplate
Native integration with modern developer toolchains

Production Caveats & Gaps

Priced above utility flash models, making it less ideal for raw grep tasks
Requires explicit instruction to format diffs as unified patches

Primary Sources & Methodology Citations

OpenAI GPT-5 Technical Architecture — OpenAI (August 2025)

Head-to-Head Frontier Comparisons: GPT 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
GPT 5OpenAICurrent	Tier 1: Master Autonomous Architect	69.5%	74%	1418	91.2%	$2.5 / $10
Claude 3.7 SonnetAnthropic Compare GPT 5 vs Claude 3.7 Sonnet →	Tier 1: Frontier Architect & Planner	70.3%	59.2%	1390	80%	$3 / $15
Gemini 3.1 Pro PreviewGoogle Compare GPT 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 GPT 5 vs xAI Grok 4.20 →	Tier 1: Frontier Systems Specialist	58.5%	54.2%	1375	76%	$2 / $8
OpenAI o3OpenAI Compare GPT 5 vs OpenAI o3 →	Tier 1: Frontier Architect & Deductive Specialist	71.7%	76.2%	1415	92.4%	$5 / $20

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

GPT 5 model overview

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

Provider: OpenAI
Family: Gpt 5
Context window: 400K tokens
Thinking support: low · medium · high
Availability status: Available
Catalog ID: openai/gpt-5

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

Run GPT 5 in Swarm

Verified CLI workflow & multi-agent architecture

Quickstart Guide →

bash — swarm run

$swarm ctl model set --provider openai --model openai/gpt-5

# Or switch interactively inside the terminal TUI:

>/models

Subagent Role Matrix

Assign GPT 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 →