Claude Opus 4.6 vs GPT-5.2 Codex
Wins 3 of 8 areas
Agents · Reasoning · Facts
Wins 2 of 8 areas
Long documents · Following instructions
Claude Opus 4.6 wins more areas, narrowly.GPT-5.2 Codex is cheaper and better at following instructions.
Scores updated · 28 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking30Claude Opus 4.63 of 4 tests · 1 tie
- FactsGetting facts right instead of making them up20Claude Opus 4.62 of 2 tests
- AgentsCarrying out multi-step tasks on its own10Claude Opus 4.61 of 1 test
- Following instructionsDoing exactly what it is asked02GPT-5.2 Codex2 of 2 tests
- Long documentsFinding answers in very long texts01GPT-5.2 Codex1 of 1 test
- CodingWriting and fixing software33Even3 each · 1 tie
- Images and chartsUnderstanding pictures, charts and video00Even0 each · 1 tie
- MathCompetition and research-level math00Even0 each · 1 tie
Agents, images and charts, math and long documents 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.
GPT-5.2 Codex costs 48% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Claude Opus 4.6 pulls ahead
- Real work tasks from 44 professionsGDPVal+16.5points ahead
- Fresh logic puzzles that change regularlyLiveBench · Reasoning+11points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+10.6points ahead
Where GPT-5.2 Codex pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+24.5points ahead
- Programming problems, refreshed regularlyLiveBench · Coding+5.4points ahead
- Reasons across sets of long documentsAA-LCR+4.3points ahead
Every test, side by side
All 28 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingEven
- LMArena · WebDevClaude Opus 4.6 by 207 rating points15461339+207 rating
- Terminal-Bench HardClaude Opus 4.6 by 9.146.237.1+9.1
- SWE-bench VerifiedClaude Opus 4.6 by 880.872.8+8
- LiveBench · CodingGPT-5.2 Codex by 5.478.283.6+5.4
- SWE-bench ProGPT-5.2 Codex by 353.456.4+3
- SciCodeGPT-5.2 Codex by 2.751.954.6+2.7
- LiveBench · Agentic Codingtie4949.4tie
ReasoningClaude Opus 4.6
- LiveBench · ReasoningClaude Opus 4.6 by 1188.777.7+11
- Humanity's Last ExamClaude Opus 4.6 by 4.239.935.7+4.2
- CritPtClaude Opus 4.6 by 3.912.68.7+3.9
- GPQA Diamondtie89.689.9tie
FactsClaude Opus 4.6
- AA-Omniscience · Non-hallucinationClaude Opus 4.6 by 10.637.226.6+10.6
- AA-Omniscience · AccuracyClaude Opus 4.6 by 5.94741.1+5.9
Following instructionsGPT-5.2 Codex
- IFBenchGPT-5.2 Codex by 24.553.177.6+24.5
- LiveBench · Instruction FollowingGPT-5.2 Codex by 3.263.366.5+3.2
Other results9 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.
- AA-OmniscienceClaude Opus 4.6 by 15.913.7-2.2+15.9
- AA Agentic IndexClaude Opus 4.6 by 11.167.656.5+11.1
- livebench_languageClaude Opus 4.6 by 9.683.373.7+9.6
- livebench_data_analysisGPT-5.2 Codex by 8.369.978.2+8.3
- Artificial Analysis Coding IndexClaude Opus 4.6 by 5.148.143+5.1
- AA IntelligenceClaude Opus 4.6 by 3.431.928.5+3.4
- LiveBenchClaude Opus 4.6 by 276.374.3+2
- Terminal-Bench 2.0Claude Opus 4.6 by 1.465.464+1.4
- τ²-Bench Telecom (AA run)tie92.192.1tie
Questions people ask
Which is better, Claude Opus 4.6 or GPT-5.2 Codex?
Claude Opus 4.6 wins three of the eight areas we test: agents, reasoning and facts. GPT-5.2 Codex wins long documents and following instructions, and costs 48% less. They are level on coding, images and charts and math.
Which is better for coding?
Neither. They win 3 coding tests each of the 7 both models report, and 1 is a tie.
Which is cheaper?
Claude Opus 4.6 costs $5.00 per million input tokens and $25.00 per million output tokens; GPT-5.2 Codex costs $1.75 and $14.00. That makes GPT-5.2 Codex about 48% cheaper for the same work.
How do you compare the two?
We use the 28 benchmark tests both models have published scores on. The verdict counts the 19 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 9 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.