Skip to content
VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Models741Deals30d271Policy30d67, 15 pendingSources1,014

GPT-5.5

OpenAIReleased Apr 23, 2026Current

Score basis

GPT-5.5 is strong in long context and reasoning; and capable in coding, math, agentic tasks, factuality, multimodal tasks, and instruction following. Too few results yet to rate safety or multilingual tasks.

Capability profile

Bars show the model's median result as a share of the leading model's, per capability. Select a row to see its results.

Strong2 capabilities
  1. Long Context−4.8%1 of 3
  2. Reasoning−6.0%6 of 6
Capable6 capabilities
  1. Coding−13.0%7 of 10
  2. Math−13.2%5 of 5
  3. Agentic−13.5%7 of 7
  4. Factuality−18.0%4 of 4
  5. Multimodal−19.9%2 of 6
  6. Instruction Following−20.1%2 of 3
Not rated2 capabilities

Too few results yet to rate Safety or Multilingual.

Price

$5.00input$30.00outputper million tokens

From Artificial Analysis · 2 providers tracked · All prices

Costs more than 94% of 330 priced models · 3:1 input-to-output blend, log scale

Evidence

169results on130benchmarks

  • 29 independently verified
  • 40 aggregator
  • 30 vendor-reported
  • 70 cross-referenced

From 34 sources · latest Oct 8, 2026 · How verification works

API features

Tool callingStructured outputsReasoning

As listed by OpenRouter

Research

224 papers reference GPT-5.5

GPT-5.5 benchmark results

169 results on 130 benchmarks, grouped by capability. Each bar is the result as a share of the capability leader's; every result links to its source.

Long Context

StrongFull long context ranking

4.8% behind the leader1 of 3 ranked benchmarks measured

Show 1 more long context resultHide 1 long context result

Reasoning

StrongFull reasoning ranking

6.0% behind the leader6 of 6 ranked benchmarks measured

Show 5 more reasoning resultsHide 5 reasoning results

Coding

CapableFull coding ranking

13.0% behind the leader7 of 10 ranked benchmarks measured

Show 7 more coding resultsHide 7 coding results

Math

CapableFull math ranking

13.2% behind the leader5 of 5 ranked benchmarks measured

Show 5 more math resultsHide 5 math results

Agentic

CapableFull agentic ranking

13.5% behind the leader7 of 7 ranked benchmarks measured

Show 9 more agentic resultsHide 9 agentic results

Factuality

CapableFull factuality ranking

18.0% behind the leader4 of 4 ranked benchmarks measured

Show 2 more factuality resultsHide 2 factuality results

Multimodal

CapableFull multimodal ranking

19.9% behind the leader2 of 6 ranked benchmarks measured

Show 5 more multimodal resultsHide 5 multimodal results

Instruction Following

CapableFull instruction following ranking

20.1% behind the leader2 of 3 ranked benchmarks measured

Show 1 more instruction following resultHide 1 instruction following result

More results

Benchmarks outside the capability baskets. They are not ranked against a leader.

Show 97 more resultsHide 97 results

GPT-5.5: common questions

Who makes GPT-5.5?

GPT-5.5 is made by OpenAI.

When was GPT-5.5 released?

GPT-5.5 was released on Apr 23, 2026, according to Artificial Analysis.

What is GPT-5.5 good at?

GPT-5.5 is strong in long context and reasoning; and capable in coding, math, agentic tasks, factuality, multimodal tasks, and instruction following. Too few results yet to rate safety or multilingual tasks.

How much does GPT-5.5 cost?

GPT-5.5 costs $5.00 per million input tokens and $30.00 per million output tokens, according to Artificial Analysis. We track its price at 2 providers. At a mix of three input tokens to one output token, it costs more than 94% of the 330 priced models we track.

How many benchmarks has GPT-5.5 been tested on?

We track 169 results for GPT-5.5 on 130 benchmarks from 34 sources, 29 of them independently verified. The latest was recorded on Oct 8, 2026.

Which API features does GPT-5.5 support?

OpenRouter lists tool calling, structured outputs, and reasoning for GPT-5.5.

About this record

Where GPT-5.5's numbers come from, and every name it appears under.

Tracked since
Apr 27, 2026
Newest source mention
Aug 24, 2026

Where the results come from

Verification: 169 scores · 29 independently verified · 40 aggregator-attributed · 70 vendor cross-reference · 30 vendor-reported. How these tiers are assigned

From 34 sources on 18 sites. Artificial Analysis supplies 40 of them; the 29 independently verified results come from 9 sites. Bars are coloured by trust tier.

  • artificialanalysis.ai40
  • lf3-static.bytednsdoc.com28
  • huggingface.co21
  • api.llm-stats.com16
  • www-cdn.anthropic.com15
  • deploymentsafety.openai.com11
  • livebench.ai7
  • ai.meta.com4
  • datasets-server.huggingface.co4
  • epoch.ai4
  • matharena.ai4
  • raw.githubusercontent.com4
  • openai.com3
  • arcprize.org2
  • labs.scale.com2
  • thinkingmachines.ai2
  • lmarena.ai1
  • simple-bench.com1

Also known as

How our sources name GPT-5.5 at each reasoning setting.

SettingShort formAPI id
low—gpt-5-5-low gpt-5.5 (low)
medium—gpt-5-5-medium gpt-5.5 (medium)
high—gpt-5-5-high gpt-5.5 (high) gpt-5.5-high gpt-5.5-high (codex-harness)
xhighgpt-5.5 xhighgpt-5.5 (xhigh) gpt-5.5-xhigh (codex-harness)
Also listed asgpt-5-5gpt-5-5-non-reasoninggpt-5.5 (codex-harness)gpt-5.5 (non-reasoning)gpt-5.5 (reasoning)gpt-5.5 thinkinggpt-5.5:batchgpt-5.5-2026-04-23