Claude 3.5 Haiku vs GPT-5 Codex
Wins 0 of 7 areas
—
Wins 6 of 7 areas
Coding · Agents · Reasoning · Images and charts · Long documents · Following instructions
GPT-5 Codex is the stronger all-rounder.
Scores updated · 17 tests both models report · How we compare
Where each one wins
Tests won in each of the seven areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- CodingWriting and fixing software03GPT-5 Codex3 of 3 tests
- ReasoningHard problems that need careful thinking03GPT-5 Codex3 of 3 tests
- AgentsCarrying out multi-step tasks on its own01GPT-5 Codex1 of 1 test
- Images and chartsUnderstanding pictures, charts and video01GPT-5 Codex1 of 1 test
- Long documentsFinding answers in very long texts01GPT-5 Codex1 of 1 test
- Following instructionsDoing exactly what it is asked01GPT-5 Codex1 of 1 test
- FactsGetting facts right instead of making them up11Even1 each
Agents, images and charts, long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where GPT-5 Codex pulls ahead
- Reasons across sets of long documentsAA-LCR+44.4points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+42.9points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+35.6points ahead
Where Claude 3.5 Haiku pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+34.3points ahead
Every test, side by side
All 17 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5 Codex
- Terminal-Bench HardGPT-5 Codex by 35.62.337.9+35.6
- SWE-bench VerifiedGPT-5 Codex by 33.940.674.5+33.9
- SciCodeGPT-5 Codex by 13.527.440.9+13.5
ReasoningGPT-5 Codex
- GPQA DiamondGPT-5 Codex by 42.940.883.7+42.9
- Humanity's Last ExamGPT-5 Codex by 24.23.627.8+24.2
- CritPtGPT-5 Codex by 5.105.1+5.1
FactsEven
- AA-Omniscience · Non-hallucinationClaude 3.5 Haiku by 34.358.924.6+34.3
- AA-Omniscience · AccuracyGPT-5 Codex by 25.513.238.7+25.5
Following instructionsGPT-5 Codex
- IFBenchGPT-5 Codex by 31.342.874.1+31.3
Other results5 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- τ²-Bench Telecom (AA run)GPT-5 Codex by 62.224.686.8+62.2
- AA Agentic IndexGPT-5 Codex by 50.81.952.7+50.8
- Artificial Analysis Coding IndexGPT-5 Codex by 2315.938.9+23
- AA IntelligenceGPT-5 Codex by 168.924.9+16
- AA-OmniscienceGPT-5 Codex by 15-22.5-7.5+15
Questions people ask
Which is better, Claude 3.5 Haiku or GPT-5 Codex?
GPT-5 Codex wins six of the seven areas where both have results: coding, agents, reasoning, images and charts, long documents and following instructions. Claude 3.5 Haiku wins none. They are level on facts.
Which is better for coding?
GPT-5 Codex. It wins 3 of the 3 coding tests both models report; Claude 3.5 Haiku wins none.
How do you compare the two?
We use the 17 benchmark tests both models have published scores on. The verdict counts the 12 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 5 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.