Claude Sonnet 5 vs GPT-5.6 Terra
Wins 1 of 8 areas
Coding
Wins 5 of 8 areas
Agents · Reasoning · Facts · Math · Long documents
GPT-5.6 Terra is the stronger all-rounder.Claude Sonnet 5 is cheaper and better at coding.
Scores updated · 46 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 thinking14GPT-5.6 Terra4 of 5 tests
- AgentsCarrying out multi-step tasks on its own03GPT-5.6 Terra3 of 4 tests · 1 tie
- MathCompetition and research-level math03GPT-5.6 Terra3 of 3 tests
- Long documentsFinding answers in very long texts02GPT-5.6 Terra2 of 2 tests
- FactsGetting facts right instead of making them up12GPT-5.6 Terra2 of 3 tests
- CodingWriting and fixing software31Claude Sonnet 53 of 6 tests · 2 ties
- Images and chartsUnderstanding pictures, charts and video11Even1 each · 1 tie
- Following instructionsDoing exactly what it is asked00Even0 each · 1 tie
Following instructions 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.
Claude Sonnet 5 costs 14% 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 Terra pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+39points ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+21.3points ahead
- Unpublished advanced math problemsFrontierMath Tiers 1-3 (v2)+20.4points ahead
Where Claude Sonnet 5 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+48.5points ahead
- Common-sense trick questionsSimpleBench+11.7points ahead
- Fixes real GitHub issues, working as an agentLiveBench · Agentic Coding+4.4points ahead
Every test, side by side
All 46 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Sonnet 5
- Terminal-Bench 2.1GPT-5.6 Terra by 7.580.588+7.5
- LiveBench · Agentic CodingClaude Sonnet 5 by 4.459.455+4.4
- LiveBench · CodingClaude Sonnet 5 by 2.480.778.3+2.4
- LMArena · WebDevClaude Sonnet 5 by 19 rating points15411522+19 rating
- SciCodetie54.355tie
- SWE-bench Protie63.263.4tie
AgentsGPT-5.6 Terra
- Terminal-Bench 4.0GPT-5.6 Terra by 21.314.135.4+21.3
- τ-Bench V3 · BankingGPT-5.6 Terra by 2.937.340.2+2.9
- BrowseCompGPT-5.6 Terra by 2.884.787.5+2.8
- GDPValtie48.347.7tie
ReasoningGPT-5.6 Terra
- CritPtGPT-5.6 Terra by 13.116.930+13.1
- SimpleBenchClaude Sonnet 5 by 11.760.648.9+11.7
- LiveBench · ReasoningGPT-5.6 Terra by 1.988.790.6+1.9
- Humanity's Last ExamGPT-5.6 Terra by 1.641.342.9+1.6
- GPQA DiamondGPT-5.6 Terra by 1.491.192.5+1.4
FactsGPT-5.6 Terra
- AA-Omniscience · Non-hallucinationClaude Sonnet 5 by 48.560.612.1+48.5
- SimpleQA VerifiedGPT-5.6 Terra by 9.533.743.2+9.5
- AA-Omniscience · AccuracyGPT-5.6 Terra by 6.74046.8+6.7
Images and chartsEven
- MMMU-ProGPT-5.6 Terra by 3.477.380.7+3.4
- CharXiv (RQ)Claude Sonnet 5 by 2.488.385.9+2.4
- LMArena · Visiontie12751271tie
MathGPT-5.6 Terra
- FrontierMath Tier 4GPT-5.6 Terra by 3929.368.3+39
- FrontierMath Tiers 1-3 (v2)GPT-5.6 Terra by 20.465.686+20.4
- LiveBench · MathematicsGPT-5.6 Terra by 292.994.9+2
Long documentsGPT-5.6 Terra
- MRCR v2 8 needle 128k (average)GPT-5.6 Terra by 1281.593.5+12
- AA-LCRGPT-5.6 Terra by 18283+1
Following instructionsEven
- LiveBench · Instruction Followingtie63.964.6tie
Other results19 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.
- GDP (Surge AI)Claude Sonnet 5 by 56.981.624.7+56.9
- HLE-VerifiedGPT-5.6 Terra by 20.13151.1+20.1
- AA-OmniscienceClaude Sonnet 5 by 16.416.40.1+16.4
- DeepSWE 1.1GPT-5.6 Terra by 165470+16
- LVBenchGPT-5.6 Terra by 10.468.578.9+10.4
- Terminal-Bench 4.0GPT-5.6 Terra by 9.112.421.5+9.1
- livebench_languageGPT-5.6 Terra by 7.97582.9+7.9
- livebench_data_analysisGPT-5.6 Terra by 7.671.779.3+7.6
- OSWorld 2.0 (partial)GPT-5.6 Terra by 7.642.650.2+7.6
- Terminal-bench 3.0GPT-5.6 Terra by 6.214.620.8+6.2
- GDPval-AA v2 EloClaude Sonnet 5 by 56 rating points15841528+56 rating
- Agent's Last Exam (Pass rate)Claude Sonnet 5 by 5.333.328+5.3
- Artificial Analysis Coding IndexGPT-5.6 Terra by 5.271.576.7+5.2
- AA IntelligenceGPT-5.6 Terra by 3.938.242.1+3.9
- Code Arena (Elo)Claude Sonnet 5 by 18 rating points15411523+18 rating
- ToolathlonClaude Sonnet 5 by 1.254.353.1+1.2
- GDP.PDF (All pass rate)GPT-5.6 Terra by 12829+1
- AA Agentic Indextie44.343.7tie
- HealthBench Professionaltie57.857.7tie
Questions people ask
Which is better, Claude Sonnet 5 or GPT-5.6 Terra?
GPT-5.6 Terra wins five of the eight areas we test: agents, reasoning, facts, math and long documents. Claude Sonnet 5 wins coding, and costs 14% less. They are level on images and charts and following instructions.
Which is better for coding?
Claude Sonnet 5. It wins 3 of the 6 coding tests both models report; GPT-5.6 Terra wins 1, and 2 are ties.
Which is cheaper?
Claude Sonnet 5 costs $2.00 per million input tokens and $10.00 per million output tokens; GPT-5.6 Terra costs $2.00 and $12.00. That makes Claude Sonnet 5 about 14% cheaper for the same work.
How do you compare the two?
We use the 46 benchmark tests both models have published scores on. The verdict counts the 27 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 19 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.