BenchLeader

GPT-5 vs GPT-5.5 Instant

Full profiles: GPT-5 (OpenAI) and GPT-5.5 Instant (OpenAI): every benchmark, price and rank.

Updated · Data from 9 sources

As of 16 Nov 2026, GPT-5 and GPT-5.5 Instant are neck and neck on quality (59.0 vs 56.9 on the BenchLeader Index, inside the ±8.1 margin), and GPT-5 is 3.3× cheaper at $3.44 vs $11.25 per 1M tokens blended. Choose GPT-5 for maths (65 vs 48), long context (70 vs 59), and agents & tools (58 vs 50); choose GPT-5.5 Instant for fast first answers (1.8 s vs 22.9 s) and reasoning (54 vs 48), coding (61 vs 57), and multimodal (58 vs 57), where it also streams ~1.5× faster.

Key facts (as of 16 Nov 2026)

  • GPT-5 scores 59.0 on the BenchLeader Index and GPT-5.5 Instant scores 56.9 (BenchLeader, 16 Nov 2026).
  • GPT-5 costs $3.44 per 1M tokens blended (list price, 16 Nov 2026).
  • GPT-5.5 Instant costs $11.25 per 1M tokens blended (list price, 16 Nov 2026).
  • GPT-5.5 Instant streams ~1.5× faster than GPT-5 (Artificial Analysis, 7-day median, 16 Nov 2026).

Which to pick for…

The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.

Use casePickWhyConfidence
CodingGPT-5.5 Instant61 vs 57 category scoreMedium
Agents & tool useGPT-558 vs 50 category score; Terminal-Bench Hard 37.9% vs 42.4%High
MathsGPT-565 vs 48 category score; OTIS Mock AIME 87.2% vs 68.1%High
ReasoningGPT-5.5 Instant54 vs 48 category score; GPQA Diamond 82.5% vs 85.3%Medium
Long documentsGPT-570 vs 59 category score; AA-LCR v1.1 76.0% vs 70.0%High
Images and other mediaGPT-5.5 Instant58 vs 57 category scoreMedium
Following instructionsEither69 vs 70 category score; IFBench 70.6% vs 71.5%Either
Chat quality (human preference)–Only GPT-5.5 Instant has an Arena rating (66 category score)Not comparable
Lowest costGPT-5$3.44 vs $11.25 per 1M blended; agentic coding session $0.115 vs $0.420–
Fastest first answerGPT-5.5 Instant1.8 s (GPT-5.5 Instant) vs 22.9 s (GPT-5 (medium)) to the first answer token–
High volume on a budgetNeither flagshipA smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs–

Key differences

  • GPT-5 (medium) and GPT-5.5 Instant are neck and neck on quality (59.0 vs 56.9, within the ±8.1 margin).
  • GPT-5 (medium) is stronger in agents & tools, long context, maths.
  • GPT-5.5 Instant is stronger in coding, instruction following, multimodal, reasoning.
  • GPT-5 (medium) is 3.3× cheaper ($3.44 vs $11.25 per 1M blended).
  • GPT-5.5 Instant streams 1.5× faster (125 vs 82 tokens per second).

At a glance

SpecGPT-5GPT-5.5 Instant
DeveloperOpenAIOpenAI
Released7 Aug 20255 May 2026
Context window (tokens)400,000400,000
Compared atGPT-5 (medium)GPT-5.5 Instant
Open weightsNoNo
Price in / out / cached input per 1M$1.25 / $10.00 / $0.125$5.00 / $30.00 / $0.500

Category by category

Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.

CategoryGPT-5 (medium)GPT-5.5 InstantLeader
Reasoning47.953.6GPT-5.5 Instant
Coding57.460.9GPT-5.5 Instant
Agents & tools58.149.9GPT-5
Maths65.448.3GPT-5
Instruction following69.470.2level
Human preference–66.1not comparable
Multimodal56.658.2GPT-5.5 Instant
Long context70.459.4GPT-5

On one chart

Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.

2550 avg75ReasoningCodingAgents & toolsMathsInstructionfollowingLong contextMultimodalGPT-5, Reasoning: 47.9GPT-5, Coding: 57.4GPT-5, Agents & tools: 58.1GPT-5, Maths: 65.4GPT-5, Instruction following: 69.4GPT-5, Long context: 70.4GPT-5, Multimodal: 56.6GPT-5.5 Instant, Reasoning: 53.6GPT-5.5 Instant, Coding: 60.9GPT-5.5 Instant, Agents & tools: 49.9GPT-5.5 Instant, Maths: 48.3GPT-5.5 Instant, Instruction following: 70.2GPT-5.5 Instant, Long context: 59.4GPT-5.5 Instant, Multimodal: 58.2

Not drawn, measured for fewer than two of these models: Knowledge, Human preference.

Model, click to add or removeIndex

Where sources disagree

Across the 10 benchmarks both have, GPT-5 leads 5, GPT-5.5 Instant leads 5. Every overall index that rates both puts GPT-5.5 Instant ahead.

Overall indexGPT-5 (medium)GPT-5.5 InstantLeader
BenchLeader Index59.056.9neck and neck
Arena Text1435†1473GPT-5.5 Instant

Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).

Price, speed and latency

List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.

MeasureGPT-5 (medium)GPT-5.5 Instant
Blended price per 1M tokens$3.44$11.25
Input / output per 1M$1.25 / $10.00$5.00 / $30.00
Cached input per 1M$0.125$0.500
Price per task: chat reply$0.0035$0.011
Price per task: summarize a 30-page report$0.021$0.078
Price per task: code edit$0.022$0.075
Price per task: agentic coding session$0.115$0.420
Price per task: structured extraction$0.0045$0.016
Agentic coding session, input mostly cached$0.064$0.217
Output speed82 tok/s125 tok/s
Time to first answer22.9 s1.8 s
Time to a complete answer, last measured 13 Nov 202666.2 s47.7 s

Each model is compared at its best-ranked reasoning-effort setting. GPT-5 (medium) scores 59.0 and takes 23 s to its first answer token; GPT-5 (high) scores 58.2 in 49 s, GPT-5 scores 56.6 in 5 s. GPT-5.5 Instant scores 56.9 and takes 2 s to its first answer token. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.

Every benchmark

122 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.

MetricPublisherGPT-5 (medium)GPT-5.5 Instant
BenchLeader IndexBenchLeader59.056.9
Agents & tools scoreBenchLeader58.149.9
Coding scoreBenchLeader57.460.9
Instruction following scoreBenchLeader69.470.2
Long context scoreBenchLeader70.459.4
Maths scoreBenchLeader65.448.3
Multimodal scoreBenchLeader56.658.2
Reasoning scoreBenchLeader47.953.6
Human preference scoreBenchLeader–66.1
Blended price $/MBenchLeader$3.44$11.25
Output speedBenchLeader82 tok/s125 tok/s
Time to first answerBenchLeader22.9 s1.8 s
Context windowBenchLeader400,000 tokens400,000 tokens
Reasoning
GPQA DiamondEpoch AI Benchmarking Hub85.3%82.5%
Humanity's Last ExamScale AI / CAIS25.3%†–
SimpleBenchSimpleBench56.7%†–
Arena Hard PromptsArena (LMArena)1449†1491
GPQA Diamond (AA)Artificial Analysis84.2%84.7%
Humanity's Last Exam (AA)Artificial Analysis25.4%21.6%
HLE-Diamond (AA)Artificial Analysis–14.6%
GPQA Diamond (Vals)Vals AI85.6%†–
Kagi LLM BenchmarkKagi LLM Benchmark72.7%†–
GPQA Diamond (HELM)HELM Capabilities79.1%†–
ARC-AGI-1ARC Prize56.2%–
ARC-AGI-2ARC Prize7.5%–
CritPtArtificial Analysis0.0%2.6%
CritPt v2 hard subset (AA)Artificial Analysis11.5%†–
Arena ExpertArena (LMArena)1445†1474
Chess PuzzlesEpoch AI Benchmarking Hub32.0%–
Mystery Game PuzzlesEpoch AI Benchmarking Hub15.0%–
LMCAEpoch AI Benchmarking Hub40.0%†–
DTBenchEpoch AI Benchmarking Hub90.7%†–
ForecastBenchForecastBench65.0%†–
Coding
SWE-bench Verified (Epoch)Epoch AI Benchmarking Hub71.5%–
SciCodeSciCode42.9%†52.5%
WeirdMLWeirdML39.8%†–
GSO-BenchGSO-Bench6.9%†–
Arena CodingArena (LMArena)1462†1513
Show all 122 benchmarks
MetricPublisherGPT-5 (medium)GPT-5.5 Instant
Coding
Arena WebDevArena (LMArena)1417–
SciCode (AA)Artificial Analysis–52.5%
LiveCodeBenchVals AI85.9%†–
SWE-bench (Vals)Vals AI69.0%†–
SWE-Bench ProScale AI SEAL41.8%†–
Aider PolyglotAider polyglot leaderboard86.7%–
SWE-bench Verified (bash only)SWE-bench65.0%–
SWE-bench Verified (any scaffold)SWE-bench74.4%†–
Vibe Code Bench v1.1Vals AI20.1%†–
SWE-Bench Pro (private)Scale AI SEAL14.9%†–
MirrorCodeEpoch AI Benchmarking Hub4.4%†–
ALE-BenchEpoch AI Benchmarking Hub1162.5†–
AlgoTuneAlgoTune1.7†–
Terminal-Bench 4.0 (AA)Artificial Analysis–12.6%
Terminal-Bench 2.1 (AA)Artificial Analysis35.2%†34.8%
GameBench 2 by reasoning settingDuelLab GameBench 238.0–
Design Arena WebsiteDesign Arena1174†–
LiveCodeBench (AA)Artificial Analysis70.3%–
Agents & tools
Terminal-BenchTerminal-Bench49.6%–
GDPvalOpenAI34.8%–
Remote Labor IndexScale AI / CAIS1.7%†–
Terminal-Bench HardArtificial Analysis37.9%42.4%
τ²-Bench Telecom (AA)Artificial Analysis86.5%49.4%
GDPval-AA v2.1, normalised scoreArtificial Analysis–0.4%
GDPval-AA v2.1Artificial Analysis925†516
τ³-Banking (AA)Artificial Analysis22.1%†11.8%
Poker AgentVals AI1103.2%†–
Terminal-Bench 2.0 (Vals)Vals AI37.1%†–
Arena SearchArena (LMArena)1189†–
BALROGBALROG32.8%†–
DeepResearch BenchDeepResearch Bench48.6%–
METR Time HorizonsMETR69.6%–
AutomationBenchArtificial Analysis–36.4%
GDP.pdf (AA)Artificial Analysis–20.2%
AA-Briefcase v1.1Artificial Analysis–1086
BrowseComp-PlusBrowseComp-Plus57.6%†–
BrowseCompOpenAI20.1%†–
DeepSearchQAGoogle59.4%†–
Maths
FrontierMath Tiers 1–3Epoch AI Benchmarking Hub55.4%†26.3%
FrontierMath Tier 4Epoch AI Benchmarking Hub21.9%†2.4%
OTIS Mock AIMEEpoch AI Benchmarking Hub87.2%68.1%
MATH Level 5Epoch AI Benchmarking Hub97.9%–
ProofBenchVals AI18.0%†–
AIME (Vals)Vals AI93.4%†–
MGSMVals AI92.8%†–
IMO 2025MathArena38.1%†–
MathArena ApexMathArena1.0%†–
Omni-MATH (HELM)HELM Capabilities64.7%†–
MATH 500Vals AI96.0%†–
Arena MathsArena (LMArena)1411†1462
AIME 2025 (AA)Artificial Analysis91.7%–
Knowledge
SimpleQA VerifiedEpoch AI Benchmarking Hub50.1%†–
MMLU-ProVals AI86.5%†–
LegalBenchVals AI86.0%†–
CorpFinVals AI61.1%†–
TaxEvalVals AI73.4%†–
MedQAVals AI96.3%†–
PRBench FinanceScale AI SEAL51.3%†–
PRBench LegalScale AI SEAL49.0%†–
MultiNRCScale AI SEAL52.1%†–
MMLU-Pro (HELM)HELM Capabilities86.3%†–
CaseLaw v2Vals AI66.5%†–
MedCodeVals AI49.6%†–
MedScribeVals AI83.7%†–
EBR-benchEpoch AI Benchmarking Hub12.7%†–
Global-MMLU-Lite (AA)Artificial Analysis90.4%–
Instruction following
IFBenchArtificial Analysis70.6%71.5%
MultiChallengeScale AI SEAL63.2%†–
TutorBenchScale AI SEAL55.3%†–
IFEval (HELM)HELM Capabilities87.5%†–
WildBench (HELM)HELM Capabilities85.7%†–
Arena Instruction FollowingArena (LMArena)1415†1456
Human preference
Arena TextArena (LMArena)1427†1473
EQ-Bench Creative Writing v3EQ-Bench1626†–
Arena Creative WritingArena (LMArena)1389†1454
Arena Multi-turnArena (LMArena)1449†1483
Arena Longer QueriesArena (LMArena)1435†1475
Lech Mazur Creative WritingLech Mazur86.0%–
Multimodal
Arena VisionArena (LMArena)1232†1251
MMMU-ProArtificial Analysis74.3%–
VISTAScale AI SEAL49.7%†–
MMMU-Pro (Vals)Vals AI81.5%†–
MortgageTaxVals AI65.5%†–
VTBScale AI SEAL17.0%†–
GeoBenchGeoBench81.0%–
VPCTEpoch AI Benchmarking Hub63.2%–
Long context
Fiction.LiveBench 120kFiction.live96.9%–
AA-LCR v1.1Artificial Analysis76.0%70.0%
Arena DocumentArena (LMArena)–1460
MLCRArtificial Analysis–31.7%
Composite
Epoch Capabilities IndexEpoch AI Benchmarking Hub150†–
AA Intelligence Index v4.3.2Artificial Analysis22.926.0
AA Progress Index v1Artificial Analysis264.4282.1
HELM Capabilities meanHELM Capabilities80.7%†–
Safety & honesty
FORTRESSScale AI SEAL17.0%†–
MASKScale AI SEAL79.3%†–
Metadata
AA Openness IndexArtificial Analysis5.6–

Data as of 2026-11-16. Best configuration of each model; every score links to its source on the model pages.

DuelLab GameBench 2, release gb2-20261111-7b598bb13de4 (https://benchmarks.duellab.org/releases/gb2-20261111-7b598bb13de4/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.

What changed

  • GPT-5 (medium): BenchLeader Index 58.2 → 59.0.
  • GPT-5 (medium): 1 new result (Aider Polyglot).
  • GPT-5 (medium): 2 new results (Global-MMLU-Lite (AA), AIME 2025 (AA)); 1 score revision.
  • GPT-5 (medium): 1 score revision.
  • GPT-5 (medium): 1 score revision.
  • GPT-5 (medium): 1 score revision.
  • GPT-5.5 Instant: 1 new result (AA-Briefcase v1.1).
  • GPT-5.5 Instant: BenchLeader Index 58.7 → 57.0.
  • GPT-5 (medium): 1 new result (AA Progress Index v1).
  • GPT-5.5 Instant: 2 new results (AA Progress Index v1, HLE-Diamond (AA)).

Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs

GPT-5 vs GPT-5.5 Instant: questions

Is GPT-5 better than GPT-5.5 Instant?
GPT-5 (medium) and GPT-5.5 Instant are neck and neck on quality (59.0 vs 56.9, within the ±8.1 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 16 Nov 2026, so check the category scores for your use.
GPT-5 vs GPT-5.5 Instant: which is cheaper, and will that change?
GPT-5 is cheaper today: $3.44 against $11.25 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
Is GPT-5 or GPT-5.5 Instant better for coding?
GPT-5.5 Instant scores higher in coding (61 vs 57 on the category index, where 50 is average).
Which is faster, GPT-5 or GPT-5.5 Instant?
GPT-5.5 Instant streams ~1.5× faster (125 vs 82 output tokens per second, Artificial Analysis 7-day median).
Is GPT-5 or GPT-5.5 Instant better for agents and tool use?
GPT-5 scores higher in agentic tasks (58 vs 50 on the category index, where 50 is average). On Terminal-Bench Hard, the top shared agents & tools benchmark, it is 37.9% for GPT-5 against 42.4% for GPT-5.5 Instant.
More questions (4)
Which is better for long documents?
GPT-5 scores higher on long-context benchmarks (70 vs 59). Both accept 400,000 tokens of context.
Which has lower latency?
GPT-5.5 Instant answers first: 1.8 s against 22.9 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
Do independent benchmarks agree on GPT-5 vs GPT-5.5 Instant?
Across the 10 benchmarks both have, GPT-5 leads 5, GPT-5.5 Instant leads 5. Every overall index that rates both puts GPT-5.5 Instant ahead.
When was GPT-5.5 Instant released?
GPT-5.5 Instant was released on 5 May 2026.