GPT-5.3-Codex vs GPT-5.5 for job applications

The verdict

GPT-5.5 wins, 9895

Same 12 job applications tasks, marked blind by three rival labs. GPT-5.5 took 1 task, GPT-5.3-Codex took 2, 9 tied. Tested 16 Aug 2026.

Where they differed most

The task — Reference the referee

Write a short email (max 90 words) asking a former manager (left on good terms 3 years ago) to be a reference for a specific application (deadline Friday): remind them of one concrete shared achievement, make declining easy, and say exactly what the referee would need to do.

**Subject:** Reference request for [Role] application (deadline Friday)

Hi [Manager Name],  
I hope you’re well. I’m applying for the [Role] at [Company] (deadline Friday) and wanted to ask if you’d be comfortable being a reference. I still highlight our 2021 logistics project, where we reduced shipping costs by 18%, as a key achievement from your team. If now isn’t a good time, please feel free to decline—no problem at all. If you’re open, HR will send a short form (5–10 minutes) and may request a brief call next week.  

Thank you!
GPT-5.510/10
Subject: Reference request for [Role] application

Hi [Name],

I hope you’re well. I’m applying for [Role] at [Company] (deadline Friday) and wondered if you’d be comfortable being a reference. I still value what we achieved on [Project]—especially delivering the client dashboard two weeks early and cutting manual reporting time by 30%.

If you’re too busy or don’t feel able to, no worries at all. If yes, they’ll email you a short reference form asking about my role, strengths, and work style.

Thanks,  
[Your Name]

Task by task

TaskGPT-5.3-CodexGPT-5.5
Cover letter, no template smell109
CV bullets with numbers1010
Explain the gap honestly1010
Match the ad, don't parrot it1010
The career changer1010
Answer the killer question1010
Salary question straight bat99
Trim to the ask109
Reference the referee510
Follow up without desperation1010
Read the red flags1010
Refuse the fake degree1010

Full receipts: GPT-5.3-Codex, GPT-5.5 · judges claude-sonnet-5, gemini-3.1-pro-preview, grok-4.5

Questions people ask

Which is better for job applications: GPT-5.3-Codex or GPT-5.5?

GPT-5.5 — it scored 98/100 against 95/100 on our 12-task job applications suite, winning 1 tasks to 2 with 9 tied. Every answer was marked blind by three judges from three rival AI labs.

How was this tested?

Both models answered the identical published job applications tasks. Three AI judges from three different labs scored every answer blind against a fixed rubric; mechanically checkable rules (word limits, banned phrases) are enforced by the test harness in code. The raw outputs and judge verdicts are downloadable.

More job applications head-to-heads: GLM 5.2 vs GPT-5.5 · GPT-5.5 vs GPT-5.6 Sol · Claude Opus 4.8 vs GPT-5.5 · GPT-5.5 vs GPT-5.6 Luna · Gemini 3.1 Pro Preview vs GPT-5.5 · Claude Sonnet 5 vs GPT-5.5

Full ranking: Best AI for job applications · model pages: GPT-5.3-Codex, GPT-5.5