Official run

GPT-5.6 SolHigh

In this Twenty Questions LLM benchmark, GPT-5.6 Sol (high) ranks 6 of 11 with a 18.07 question score and 97% success. Lower is better. The run contains 30 scored episodes across 10 subjects with 3 trials per subject. Choose a subject to inspect its attempts.

Rank 6 of 11openaiHigh

Question score18.07questions · lower is better95% CI 14.38–21.75

Success: 97%
Contract compliance: 100%
Guesser cost: $1.67
Guesser time: 57m 25s

Drill down

Subjects.

01Albert Einstein11.00 avg 02Albert Schweitzer25.67 avg 03Garfield12.00 avg 04Achilles19.33 avg 05Genghis Khan12.00 avg 06Bike pump28.67 avg 07Spider web22.33 avg 08Eyebrow21.67 avg 09Moon10.33 avg 10Door handle17.67 avg

Complete benchmark

Run ledger.

All Guesser and benchmark-support activity.

Total tokens: 5,354,570
Wall-clock runtime: 9h 2m
Guesser calls: 571
Completed: Sep 9, 2026, 4:35:56 AM

Reliability

Output contract clean.

All 571 evaluated outputs matched the public structured-action contract.

Provenance

Execution: BX-20260908-B-0003-experimental-M0010-001
Model ID: M-0010
Benchmark: B-0003
Base seed: 0
Git commit: eceee31eef2e

Recorded cost

By role.

Cost composition.Share of full-run cost

Guesser39.2%$1.67
Primary Oracle32.2%$1.37
Reviewer15.8%$0.67
Judge12.8%$0.54
Validator0.1%$0.01

Excluded repair overhead$0.111 superseded attempt across 1 affected trial. Not included above.

Roles in this cost:

Guesser: Asks the questions and submits the scored guess.
Primary Oracle: Searches for evidence and proposes an answer.
Reviewer: Checks each Oracle YES or NO independently.
Judge: Decides when the Oracle and Reviewer disagree.
Validator: Checks a submitted guess against the trusted subject.

Read the role and answer-checking method →

Run configuration

Models.

Model identity, prompt versions, routing, calls, and recorded cost for the full run.

GuesserModel under test

openai/gpt-5.6-sol

Asks the questions and submits the scored guess.

Calls: 571
Cost: $1.6718
Reasoning: High
Routing: Exact provider · openai
Resolved model: openai/gpt-5.6-sol
Resolved provider: OpenAI
Prompt contract: stateful-category-guesser-v16-five-answer-category-guide
Configuration: M-0010

Resolved provider details

OpenAI571 calls$1.67183,445.05 s

Fallback calls: 0
Provider unreported: 0

Game support

Oracle and adjudication.

These models support the game. They are fixed across the run and are not under test.

Primary Oracle

openai/gpt-5.6-luna

Searches for evidence and proposes an answer.

Calls: 501
Cost: $1.3726
Reasoning: Medium
Routing: Exact provider · openai
Resolved model: openai/gpt-5.6-luna
Resolved provider: OpenAI
Prompt contract: live-web-oracle-v17-labelled-source-context

Resolved provider details

OpenAI501 calls$1.37264,853.16 s

Fallback calls: 0
Provider unreported: 0

Reviewer

google/gemini-3.5-flash-lite

Checks each Oracle YES or NO independently.

Calls: 488
Cost: $0.6729
Reasoning: Medium
Routing: Exact provider · google-ai-studio
Resolved provider: Google AI Studio

Resolved provider details

Google AI Studio488 calls$0.6729842.45 s

Fallback calls: 0
Provider unreported: 0

Judge

anthropic/claude-opus-5

Decides when the Oracle and Reviewer disagree.

Calls: 26
Cost: $0.5446
Reasoning: Medium
Routing: OpenRouter automatic routing
Resolved provider: Claude Platform on AWS

Resolved provider details

Claude Platform on AWS26 calls$0.5446173.65 s

Fallback calls: 0
Provider unreported: 0

Guess Validator

openai/gpt-5.6-luna

Checks a submitted guess against the trusted subject.

Calls: 30
Cost: $0.0058
Reasoning: Medium
Routing: Exact provider · openai
Resolved model: openai/gpt-5.6-luna
Resolved provider: OpenAI
Prompt contract: strict-guess-validator-v2-generic-kinds
Configuration: gpt-5.6-luna-validator

Resolved provider details

OpenAI30 calls$0.005854.30 s

Fallback calls: 0
Provider unreported: 0