GPT-5 vs GPT-5.1
Wins 2 of 7 areas
Agents · Following instructions
Wins 4 of 7 areas
Coding · Reasoning · Images and charts · Long documents
GPT-5.1 wins more areas, narrowly.GPT-5 is better at agents.
Scores updated · 56 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.
- Images and chartsUnderstanding pictures, charts and video03GPT-5.13 of 3 tests
- CodingWriting and fixing software13GPT-5.13 of 6 tests · 2 ties
- ReasoningHard problems that need careful thinking12GPT-5.12 of 5 tests · 2 ties
- Long documentsFinding answers in very long texts01GPT-5.11 of 1 test
- AgentsCarrying out multi-step tasks on its own30GPT-53 of 3 tests
- Following instructionsDoing exactly what it is asked10GPT-51 of 2 tests · 1 tie
- FactsGetting facts right instead of making them up22Even2 each
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.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where GPT-5.1 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+30.3points ahead
- Command-line tasks in a real terminalTerminal-Bench 2.1+17.2points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+12.9points ahead
Where GPT-5 pulls ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+6.2points ahead
- Keeps track of context across a multi-turn chatMulti-Challenge+6.2points ahead
- Real work tasks from 44 professionsGDPVal+4.5points ahead
Every test, side by side
All 56 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGPT-5.1
- Terminal-Bench 2.1GPT-5.1 by 17.235.252.4+17.2
- Terminal-Bench HardGPT-5.1 by 12.932.645.5+12.9
- LMArena · WebDevGPT-5 by 23 rating points14181395+23 rating
- SWE-bench VerifiedGPT-5.1 by 1.474.976.3+1.4
- SciCodetie42.943.3tie
- LiveCodeBench v6tie8787tie
AgentsGPT-5
- τ-Bench V3 · BankingGPT-5 by 6.222.115.9+6.2
- GDPValGPT-5 by 4.52116.5+4.5
- BrowseCompGPT-5 by 4.154.950.8+4.1
ReasoningGPT-5.1
- ARC-AGI-2GPT-5.1 by 7.79.917.6+7.7
- SimpleBenchGPT-5 by 3.556.753.2+3.5
- GPQA DiamondGPT-5.1 by 1.985.487.3+1.9
- CritPttie5.74.9tie
- Humanity's Last Examtie28.528.5tie
FactsEven
- AA-Omniscience · Non-hallucinationGPT-5.1 by 30.317.848.1+30.3
- Vectara HHEM hallucination ratelower is betterGPT-5.1 by 315.112.1+3
- AA-Omniscience · AccuracyGPT-5 by 2.640.337.7+2.6
- SimpleQA VerifiedGPT-5 by 2.150.148+2.1
Images and chartsGPT-5.1
- LMArena · VisionGPT-5.1 by 42 rating points12081250+42 rating
- MMMU-ProGPT-5.1 by 1.374.275.5+1.3
- MMMUGPT-5.1 by 1.284.285.4+1.2
Following instructionsGPT-5
- Multi-ChallengeGPT-5 by 6.269.663.4+6.2
- IFBenchtie73.172.9tie
Other results32 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-OmniscienceGPT-5.1 by 14.1-8.75.4+14.1
- Terminal-Bench 2.0GPT-5.1 by 12.435.247.6+12.4
- Artificial Analysis Coding IndexGPT-5.1 by 11.637.849.4+11.6
- HealthBench Hard length-adjustedGPT-5 by 9.334.725.4+9.3
- vectara_avg_summary_lengthGPT-5.1 by 91.7 rating points162.7254.4+91.7 rating
- bbqGPT-5 by 8.196.888.7+8.1
- HMMT Feb. 2025GPT-5.1 by 888.396.3+8
- HLE (with tools)GPT-5.1 by 7.535.242.7+7.5
- ARC-AGI-1GPT-5.1 by 7.165.772.8+7.1
- HealthBench length-adjustedGPT-5 by 6.857.750.9+6.8
- HealthBench Professional length-adjustedGPT-5 by 6.646.239.6+6.6
- AA Agentic IndexGPT-5 by 4.926.521.6+4.9
- Tau2 airlineGPT-5.1 by 4.462.667+4.4
- τ²-Bench (Retail)GPT-5 by 3.281.177.9+3.2
- vectara_factual_consistencyGPT-5.1 by 384.987.9+3
- τ²-Bench Telecom (AA run)GPT-5 by 2.984.881.9+2.9
- HMMT Nov. 2025GPT-5.1 by 2.589.291.7+2.5
- matharena_visual_math_overallGPT-5 by 1.978.876.9+1.9
- AA IntelligenceGPT-5.1 by 1.72324.7+1.7
- AIR-Bench 2024GPT-5 by 1.587.786.2+1.5
- XSTestGPT-5.1 by 1.197.198.2+1.1
- swe_bench_bashGPT-5.1 by 16566+1
- AIME 2025tie94.694tie
- HealthBench Consensus length-adjustedtie95.695tie
- anthropic_red_teamtie99.199.5tie
- FrontierMath (overall)tie26.326.7tie
- τ²-Benchtie82.482.7tie
- MMLU-Protie87.187tie
- vectara_answer_ratetie99.9100tie
- frontiermath_tier_4_v1tie12.512.5tie
- HarmBenchtie97.697.6tie
- simple_safety_teststie99.899.8tie
Questions people ask
Which is better, GPT-5 or GPT-5.1?
GPT-5.1 wins four of the seven areas where both have results: coding, reasoning, images and charts and long documents. GPT-5 wins agents and following instructions. They are level on facts.
Which is better for coding?
GPT-5.1. It wins 3 of the 6 coding tests both models report; GPT-5 wins 1, and 2 are ties.
Which is cheaper?
GPT-5 costs $1.25 per million input tokens and $10.00 per million output tokens; GPT-5.1 costs $1.25 and $10.00. For a million tokens read plus a million written, they cost about the same.
How do you compare the two?
We use the 56 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 32 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.