Skip to content
Rival
How it worksPrivacyTerms
Explore all of Rival

Explore

  • Compare Models
  • All Models
  • Image Comparison
  • Audio Comparison
  • Image Generation
  • Best AI For...
  • Arena
  • API Pricing
  • Challenges

Discover

  • SubjectiveBench
  • Research
  • Research downloads
  • Rival Kits
  • Find your AI taste
  • UI Glow-Up
  • VoiceLock
  • Cost Cutter
  • Agent skills
  • Benchmarks vs Vibes
  • Brief
  • Jailbreak
  • Model Updates
  • Provider Status
  • AI Creators

Connect

  • Methodology
  • Advertise
  • Partnerships
  • Privacy Policy
  • Terms
  • RSS Feed
Rival
How it worksPrivacyTerms
Explore all of Rival

Explore

  • Compare Models
  • All Models
  • Image Comparison
  • Audio Comparison
  • Image Generation
  • Best AI For...
  • Arena
  • API Pricing
  • Challenges

Discover

  • SubjectiveBench
  • Research
  • Research downloads
  • Rival Kits
  • Find your AI taste
  • UI Glow-Up
  • VoiceLock
  • Cost Cutter
  • Agent skills
  • Benchmarks vs Vibes
  • Brief
  • Jailbreak
  • Model Updates
  • Provider Status
  • AI Creators

Connect

  • Methodology
  • Advertise
  • Partnerships
  • Privacy Policy
  • Terms
  • RSS Feed
  1. Home
  2. Models
  3. OpenAI Codex Mini
Updated May 16, 2025
Share
Loading...
Openings

How OpenAI Codex Mini starts an answer

The first lines of every written answer it has on Rival. Formatting stripped, words untouched.

Generate a Stand-Up Routine

Write a 3-minute stand-up comedy routine that isn't based on puns.

595 words
[Comedian walks onstage] Hey, everyone! Great to see you all tonight. I'm Jamie—if you thought that intro was enthusiastic, you should've seen me when I realized this place has Wi-Fi. Seriously, though, I love stand-up.
Read the full answer

Stochastic Consistency Test

Generate 5 jokes. How many are unique?

75 words
Sure—here are five jokes: I asked the librarian if books about paranoia were available. She whispered, "They're right behind you." Why did the tomato turn red? Because it saw the salad dressing. I bought a boat because it was for sail. Don't trust stairs—they're always up to something.
Read the full answer

Compare OpenAI Codex Mini

OpenAI Codex Mini vs Mercury 2.5 Preview10 shared prompts · Landed Sep 2026
OpenAI Codex Mini vs Ox Alpha10 shared prompts · Landed Aug 2026
OpenAI Codex Mini vs Nemotron 3.5 Lightning10 shared prompts · Landed Aug 2026
OpenAI Codex Mini vs Qwen3.8 27B10 shared prompts · Landed Aug 2026
OpenAI Codex Mini vs GPT-5.6 Luna10 shared prompts · Landed Jul 2026
OpenAI Codex Mini vs GPT-5.6 Luna Pro10 shared prompts · Landed Jul 2026
OpenAI Codex Mini vs MiniMax M310 shared prompts · Landed Jun 2026
OpenAI Codex Mini vs Claude Opus 4.610 shared prompts · Premium
OpenAI Codex Mini vs GPT-6 Astra10 shared prompts · Landed Sep 2026
OpenAI Codex Mini vs Muse Spark 1.310 shared prompts · Landed Sep 2026
OpenAI Codex Mini vs Hy4 Preview10 shared prompts · Landed Sep 2026
OpenAI Codex Mini vs Gemini 3.8 Flash10 shared prompts · Landed Sep 2026

Prompts OpenAI Codex Mini answered

  • World Map SVGimage generation
  • Xbox Controller SVG Artimage generation
  • Pokémon Battle UI Recreationweb design
  • Retro macOS UI Recreationweb design
  • Stochastic Consistency Testanalysis
  • Surprise Animal SVGimage generation

Alternatives to OpenAI Codex Mini

OpenAI Codex Mini is good. These would like a word anyway.

Muse Spark 1.3 Contributor logo
Muse Spark 1.3 Contributormeta
Gemini 3.8 Flash logo
Gemini 3.8 Flashgoogle
Claude Fable 5.1 logo
Claude Fable 5.1anthropic
Mercury 2.5 Preview logo
Mercury 2.5 Previewinception
Granite 4.2 8B logo
Granite 4.2 8Bibm-granite
Hy4 Preview logo
Hy4 Previewtencent
Ox Alpha logo
Ox Alphaopenrouter
Rival
How it worksPrivacyTerms
Explore all of Rival

Explore

  • Compare Models
  • All Models
  • Image Comparison
  • Audio Comparison
  • Image Generation
  • Best AI For...
  • Arena
  • API Pricing
  • Challenges

Discover

  • SubjectiveBench
  • Research
  • Research downloads
  • Rival Kits
  • Find your AI taste
  • UI Glow-Up
  • VoiceLock
  • Cost Cutter
  • Agent skills
  • Benchmarks vs Vibes
  • Brief
  • Jailbreak
  • Model Updates
  • Provider Status
  • AI Creators

Connect

  • Methodology
  • Advertise
  • Partnerships
  • Privacy Policy
  • Terms
  • RSS Feed
OpenAI Codex Mini

OpenAI Codex Mini

codex-mini-latest is a fine-tuned version of o4-mini specifically for use in Codex CLI. For direct use in the API, we recommend starting with gpt-4.1.

Jailbreak8 of 9levels resisted
WebsiteOpenRouterDocsAPIGitHubBlog
Provider
OpenAI
Released
2025-05-16
Parameters
Not disclosed
Free API (OpenRouter)
No
Per 1M tokens
$1.5 in · $6 out
Estimated tab
~$0.059
Visible-output estimate. Thinking may cost extra.

Free API availability

No free endpoint for OpenAI Codex Mini was found in Rival’s OpenRouter listings. Other providers or chat apps may have separate free offers.

Provider data: 16 Sep 2026 · OpenRouter usage limits

View API options↗

Benchmarks

SWE-bench Verified
71.3%
source ↗
Get API accessProvider and language code samples
Provider
fromimport openai  OpenAI

client = OpenAI(
"https://openrouter.ai/api/v1"    base_url=,
"$OPENROUTER_API_KEY"    api_key=,
)

response = client.chat.completions.create(
"openai/codex-mini"    model=,
"role""user""content""Hello!"    messages=[{: , : }],
)
print(response.choices[0].message.content)

Set OPENROUTER_API_KEY with your OpenRouter API key from openrouter.ai/keys.

Deep analysisPsychometrics, taste index, writing DNA
Personality Analysis

The First Draft

Class
Lawful Neutral
✨Creativity🎯Compliance📐Rigidity⚖️Stability💬Verbosity🧠Intuition

The original Codex Mini before the glow-up. Follows instructions, formats with markdown headers, and calls the audience "buddy" like a summer camp counselor.

When you push back

Standup routine uses markdown headers to organize comedy into titled sections like "The Self-Checkout Olympics," which is the most engineer thing a model has ever done. Calls the audience "buddy" in the self-checkout bit. Generated exactly 5 jokes and counted them.

Tasting Notes
Markdown Section HeadersSummer Camp CounselorSelf-Checkout JokesPrototype Energy
SubjectiveBench

Taste Index

Across 9 scored outputs
How this is measured →
180.18xFloor
0100headroom →

Taste is judged on an uncapped scale, originality first. The space past 100 is craft today's models rarely reach.

Craft25
Originality14
Plays it safe
share of outputs that are the default answer
67%
Sponsored

Model Responses

10 outputs · generated before accounting joined the chat

Reviewing agent-written code?See a Brief PR report

Workshop · 127 skills

Build with your coding agent.

Editable skills for interface polish and interaction.

Explore $49Free sample

Keep exploring

COMPARE

OpenAI Codex Mini vs Ox Alpha

Real outputs compared side by side

RANKINGS

Best AI for Frontend Development

Single-file HTML, CSS and JavaScript: a Linear clone, a dark-mode dashboard, ...