OpenAI model

GPT 5.1 Codex

GPT 5.1 Codex in the public OpenAI catalog.

Provider
OpenAI
Availability
Available
Model ID
gpt-5.1-codex
Snapshot
2026-10-07
Evaluation Verdict

OpenAI GPT-5.1 Codex is an autonomous coding model tuned specifically for repository refactoring, test-driven development, and multi-file codebases (72.0% SWE-bench, 1420 Coding Elo).

GPT 5.1 Codex Benchmarks & Agentic Accuracy

Verified primary research & developer preference metrics

Primary Sources Verified
SWE-bench Verified
72%
Autonomous Agent ModeOpenAI Codex System Card ↗
LiveCodeBench
75.5%
ThinkingLiveCodeBench ↗
LMSYS Coding Elo
1420Elo
Coding ArenaLMSYS Arena ↗
AIME 2024 / 2025
92%
ThinkingOpenAI Codex System Card ↗
Pricing ($ / 1M tokens)
$2.5/ $10
$0.6 cached readStandard API Pricing

What This Means in Real Software Development

  • Specialized in repository-level code synthesis with 72.0% on SWE-bench Verified.
  • Understands complex git worktree hierarchies and directory structures without hallucinations.
  • Native test-driven development loop: authors failing tests, inspects failures, and writes minimal passing patches.
  • Strong throughput ensures rapid completion of complex Coder jobs.

Swarm Multi-Agent Topology

Tier 1: Dedicated Systems CoderRole: Coder Subagent

Best Paired With: Claude 3.7 Sonnet as Lead Planner

Specialized architecture for dirty repository exploration, bug fixing, and test execution.

Git Worktree Isolation Advantage:

Dispatched into Swarm isolated worktrees to author and commit tested code patches.

What Developers & Practitioners Are Saying

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

Field Consensus

"Specialized Codex iteration designed for autonomous workspace execution and terminal tool use."

Verified Real-World Strengths

  • 72.0% SWE-bench Verified with native git patch generation
  • Deep understanding of modern build systems and CLI environments
  • High precision in multi-file refactoring

Production Caveats & Gaps

  • Requires careful prompt sandboxing to avoid unintended terminal commands
Benchmark Reality Check

What the benchmark tests: Measures autonomous CLI tool calling, patch authoring, and test execution.

Production caveat: Always execute in isolated worktrees to isolate experimental changes.

Primary Sources & Methodology Citations

GPT 5.1 Codex model overview

GPT 5.1 Codex 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
Not supported
Availability status
Available
Catalog ID
openai/gpt-5.1-codex
  • File input
  • Image input
  • Text input
  • Text output
  • Reasoning
  • Tools
  • Structured output
  • JSON

Run GPT 5.1 Codex in Swarm

Verified CLI workflow & multi-agent architecture

Quickstart Guide →
bash — swarm run
$swarm ctl model set --provider openai --model openai/gpt-5.1-codex

# Or switch interactively inside the terminal TUI:

>/models

Subagent Role Matrix

Assign GPT 5.1 Codex 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.
← All OpenAI models

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