Claude Sonnet 4.5 vs GPT-5.6 Luna
Wins 0 of 7 areas
—
Wins 7 of 7 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Long documents
GPT-5.6 Luna is the stronger all-rounder.
Scores updated · 37 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 software05GPT-5.6 Luna5 of 5 tests
- AgentsCarrying out multi-step tasks on its own05GPT-5.6 Luna5 of 5 tests
- ReasoningHard problems that need careful thinking14GPT-5.6 Luna4 of 5 tests
- MathCompetition and research-level math02GPT-5.6 Luna2 of 2 tests
- FactsGetting facts right instead of making them up12GPT-5.6 Luna2 of 3 tests
- Images and chartsUnderstanding pictures, charts and video12GPT-5.6 Luna2 of 3 tests
- Long documentsFinding answers in very long texts01GPT-5.6 Luna1 of 1 test
Long documents rests on a single test.
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-5.6 Luna costs 92% 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-5.6 Luna pulls ahead
- Finds hard-to-locate facts by browsing the webBrowseComp+63.7points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+58.2points ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+56.1points ahead
Where Claude Sonnet 4.5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+43.5points ahead
- Common-sense trick questionsSimpleBench+7.5points ahead
- Math problems shown in pictures and chartsMathVista+2.9points ahead
Every test, side by side
All 37 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.6 Luna
- Terminal-Bench 2.1GPT-5.6 Luna by 25.155.880.9+25.1
- SWE-bench ProGPT-5.6 Luna by 19.143.662.7+19.1
- SWE-bench VerifiedGPT-5.6 Luna by 14.878.293+14.8
- LMArena · WebDevGPT-5.6 Luna by 134 rating points13851519+134 rating
- SciCodeGPT-5.6 Luna by 7.945.753.6+7.9
AgentsGPT-5.6 Luna
- BrowseCompGPT-5.6 Luna by 63.719.683.3+63.7
- GDPValGPT-5.6 Luna by 27.520.548+27.5
- Terminal-Bench 4.0GPT-5.6 Luna by 11.6011.6+11.6
- OSWorld-VerifiedGPT-5.6 Luna by 11.261.472.6+11.2
- τ-Bench V3 · BankingGPT-5.6 Luna by 6.624.531.1+6.6
ReasoningGPT-5.6 Luna
- ARC-AGI-2GPT-5.6 Luna by 4613.659.6+46
- Humanity's Last ExamGPT-5.6 Luna by 21.717.839.5+21.7
- CritPtGPT-5.6 Luna by 19.51.120.6+19.5
- GPQA DiamondGPT-5.6 Luna by 7.783.491.1+7.7
- SimpleBenchClaude Sonnet 4.5 by 7.554.346.8+7.5
FactsGPT-5.6 Luna
- AA-Omniscience · Non-hallucinationClaude Sonnet 4.5 by 43.550.97.4+43.5
- SimpleQA VerifiedGPT-5.6 Luna by 10.330.741+10.3
- AA-Omniscience · AccuracyGPT-5.6 Luna by 9.832.942.7+9.8
Images and chartsGPT-5.6 Luna
- CharXiv (RQ)GPT-5.6 Luna by 15.567.282.7+15.5
- MMMU-ProGPT-5.6 Luna by 9.968.778.6+9.9
- MathVistaClaude Sonnet 4.5 by 2.979.876.9+2.9
MathGPT-5.6 Luna
- FrontierMath Tiers 1-3 (v2)GPT-5.6 Luna by 58.223.982.1+58.2
- FrontierMath Tier 4GPT-5.6 Luna by 56.12.458.5+56.1
Other results13 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.
- BrowseComp (context management)GPT-5.6 Luna by 57.926.184+57.9
- ARC-AGI-1GPT-5.6 Luna by 2763.790.7+27
- AA Agentic IndexGPT-5.6 Luna by 25.217.542.7+25.2
- HLE (with tools)GPT-5.6 Luna by 24.424.548.9+24.4
- Artificial Analysis Coding IndexGPT-5.6 Luna by 19.352.171.4+19.3
- AA IntelligenceGPT-5.6 Luna by 16.620.737.3+16.6
- ToolathlonGPT-5.6 Luna by 12.44153.4+12.4
- HealthBenchGPT-5.6 Luna by 11.644.255.8+11.6
- AA-OmniscienceClaude Sonnet 4.5 by 10.2-0.1-10.3+10.2
- SimpleVQAClaude Sonnet 4.5 by 4.457.653.2+4.4
- MMMU (val) (Pass@1)Claude Sonnet 4.5 by 2.677.875.2+2.6
- RealWorldQAGPT-5.6 Luna by 2.170.372.4+2.1
- HallusionBenchtie59.959tie
Questions people ask
Which is better, Claude Sonnet 4.5 or GPT-5.6 Luna?
GPT-5.6 Luna wins all seven areas where both have results: coding, agents, reasoning, facts, images and charts, math and long documents. Claude Sonnet 4.5 wins none.
Which is better for coding?
GPT-5.6 Luna. It wins 5 of the 5 coding tests both models report; Claude Sonnet 4.5 wins none.
Which is cheaper?
Claude Sonnet 4.5 costs $3.00 per million input tokens and $15.00 per million output tokens; GPT-5.6 Luna costs $0.20 and $1.20. That makes GPT-5.6 Luna about 92% cheaper for the same work.
How do you compare the two?
We use the 37 benchmark tests both models have published scores on. The verdict counts the 24 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 13 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.