UCP PlaygroundGPT-5.5/Models/GPT-5.5
InspectorAgentDocs
OpenAIOpenAI

GPT-5.5

OpenAI's frontier model (GPT-5.5, succeeding 5.2). The prior 5.2 revision had a notably high tool-error rate in our benchmark — nearly half of tool calls failed against MCP schemas, though turn efficiency was reasonable (~7.6 turns) when it reached checkout. 5.5 is the newer revision and is being re-benchmarked.

Avg Tokens52,623Avg Duration33.0sTurns to Checkout10.1
CheckoutCartSearchTurnsTokens11.224.168.7248
Shopping Score21/100
Weak

Scores computed from real agent sessions against live UCP-enabled stores. Not estimated — every data point is from an actual tool call, checkout attempt, and store response.

Shopping Score Breakdown

Checkout Rate40%11.2%
Cart Rate20%24.1%
Search Rate10%68.7%
Turn Efficiency15%24%
Token Efficiency15%8%
31.3% of sessions failed with errors.

Top Stores

StoreCheckout %Cart %
••••••••••••••••••••••••••••••••100%100%
••••••••••••••••••••••••••••••••85.7%85.7%
••••••••••••••••••••••••••••••••50%50%
kyliecosmetics.com50%50%
•••••••••••••••••••37.5%37.5%
••••••••••••••••••••••••••••••••33.3%33.3%
houseofparfum.nl30.4%39.1%
apneeswimwear.com25%25%
everlane.com22.2%22.2%
ucp.travel20%20%

Known Issues

High Tool Error Ratehigh

46.9% of all tool calls fail. Nearly half of MCP interactions produce errors.

Schema Compliancehigh

Struggles with MCP tool schemas — incorrect argument types, missing required fields.

Token Usage

Avg per Session52,623
Fleet Average66,736
vs Fleet−21%
Median (p50)24,426
p90129,545
Prompt / Completion split
Prompt 51,512 (98%) Completion 1,111 (2%)
Daily Average Tokens (last 30 days)
Distribution range
Min0p259,230p5024,426p7555,282p90129,545Max601,707

Test GPT-5.5 on Your Store

Run a live agent session to see how GPT-5.5 handles your store's checkout flow end-to-end.

Run Agent Session

Compare With Other Models

MetaMetaLlama 3.3 70B39.3%GoogleGoogleGemini 3.5 Flash39%AnthropicAnthropicClaude Sonnet 4.636.8%GoogleGoogleGemini 2.5 Flash26.4%xAIxAIGrok 4.326.3%AnthropicAnthropicClaude Opus 4.823.2%DeepSeekDeepSeekDeepSeek V3.222.1%GoogleGoogleGemini 3.1 Pro21.1%GoogleGoogleGemini 2.5 Pro20%OpenAIOpenAIGPT-4o15.5%OpenAIOpenAIGPT-5.511.2%xAIxAIGrok 3 Mini (Reasoning)11.1%AnthropicAnthropicClaude Sonnet 510%OpenAIOpenAIo4-mini (Reasoning)9.5%DeepSeekDeepSeekDeepSeek V4 Pro9%xAIxAIGrok 4.54.5%Moonshot AIMoonshot AIKimi K34.5%DeepSeekDeepSeekDeepSeek R1 (Reasoning)4.1%DeepSeekDeepSeekDeepSeek V4 Flash3.3%MetaMetaLlama 4 Maverick2.7%GoogleGoogleGemini 3.6 Flash2%AnthropicAnthropicClaude Opus 51.8%AlibabaAlibabaQwen 3.7 Max0%AlibabaAlibabaQwQ 32B (Reasoning)0%
Full comparison view →