GPT-5.4 vs GPT-6 Luna
Wins 3 of 8 areas
Reasoning · Facts · Following instructions
Wins 2 of 8 areas
Agents · Long documents
GPT-5.4 wins more areas, narrowly.GPT-6 Luna is cheaper and better at agents.
Scores updated · 30 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 thinking40GPT-5.44 of 5 tests · 1 tie
- FactsGetting facts right instead of making them up21GPT-5.42 of 3 tests
- Following instructionsDoing exactly what it is asked10GPT-5.41 of 1 test
- AgentsCarrying out multi-step tasks on its own01GPT-6 Luna1 of 1 test
- Long documentsFinding answers in very long texts01GPT-6 Luna1 of 1 test
- CodingWriting and fixing software22Even2 each
- Images and chartsUnderstanding pictures, charts and video11Even1 each
- MathCompetition and research-level math11Even1 each · 1 tie
Agents, 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.
GPT-6 Luna costs 97% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5.4 pulls ahead
- Follows detailed instructions exactlyLiveBench · Instruction Following+14.3points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+14points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+7.1points ahead
Where GPT-6 Luna pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+15points ahead
- Real work tasks from 44 professionsGDPVal+9.3points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+7.1points ahead
Every test, side by side
All 30 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingEven
- LMArena · WebDevGPT-6 Luna by 115 rating points14621577+115 rating
- LiveBench · Agentic CodingGPT-5.4 by 2.653.851.2+2.6
- SciCodeGPT-5.4 by 256.654.6+2
- LiveBench · CodingGPT-6 Luna by 1.577.579+1.5
ReasoningGPT-5.4
- ARC-AGI-2GPT-5.4 by 1473.359.3+14
- LiveBench · ReasoningGPT-5.4 by 6.388.181.8+6.3
- Humanity's Last ExamGPT-5.4 by 5.243.738.5+5.2
- CritPtGPT-5.4 by 423.419.4+4
- ARC-AGI-3tie0.20.1tie
FactsGPT-5.4
- AA-Omniscience · Non-hallucinationGPT-6 Luna by 158.323.3+15
- AA-Omniscience · AccuracyGPT-5.4 by 7.150.943.8+7.1
- SimpleQA VerifiedGPT-5.4 by 3.745.141.4+3.7
Images and chartsEven
- LMArena · VisionGPT-5.4 by 90 rating points13011211+90 rating
- MMMU-ProGPT-6 Luna by 1.378.479.7+1.3
MathEven
- FrontierMath Tier 4GPT-6 Luna by 7.14956.1+7.1
- LiveBench · MathematicsGPT-5.4 by 5.194.289.1+5.1
- FrontierMath Tiers 1-3 (v2)tie78.679tie
Following instructionsGPT-5.4
- LiveBench · Instruction FollowingGPT-5.4 by 14.370.255.9+14.3
Other results10 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.
- DeepSWE 1.1GPT-6 Luna by 14.65266.6+14.6
- HealthBench Professional length-adjustedGPT-6 Luna by 12.748.160.8+12.7
- livebench_languageGPT-5.4 by 8.882.673.8+8.8
- ARC-AGI-1GPT-5.4 by 793.786.7+7
- livebench_data_analysisGPT-5.4 by 5.979.373.4+5.9
- AA-OmniscienceGPT-5.4 by 5.15.80.7+5.1
- HealthBench Hard length-adjustedGPT-6 Luna by 2.329.131.4+2.3
- AA Intelligencetie3938.1tie
- HealthBench length-adjustedtie5454.5tie
- HealthBench Consensus length-adjustedtie96.395.9tie
Questions people ask
Which is better, GPT-5.4 or GPT-6 Luna?
GPT-5.4 wins three of the eight areas we test: reasoning, facts and following instructions. GPT-6 Luna wins agents and long documents, and costs 97% less. They are level on coding, images and charts and math.
Which is better for coding?
Neither. They win 2 coding tests each of the 4 both models report.
Which is cheaper?
GPT-5.4 costs $2.50 per million input tokens and $15.00 per million output tokens; GPT-6 Luna costs $0.10 and $0.50. That makes GPT-6 Luna about 97% cheaper for the same work.
How do you compare the two?
We use the 30 benchmark tests both models have published scores on. The verdict counts the 20 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 10 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.