Official run

GPT-6 AstraHigh

In this Twenty Questions LLM benchmark, GPT-6 Astra (high) ranks 1 of 11 with a 13.67 question score and 100% 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 1 of 11openaiHigh

Question score13.67questions · lower is better95% CI 12.49–14.85

Success: 100%
Contract compliance: 100%
Guesser cost: $5.27
Guesser time: 40m 43s

Drill down

Subjects.

01Albert Einstein9.67 avg 02Albert Schweitzer16.00 avg 03Garfield8.00 avg 04Achilles8.00 avg 05Genghis Khan10.00 avg 06Bike pump26.33 avg 07Spider web18.33 avg 08Eyebrow12.67 avg 09Moon10.00 avg 10Door handle17.67 avg

Complete benchmark

Run ledger.

All Guesser and benchmark-support activity.

Total tokens: 3,043,336
Wall-clock runtime: 1h 58m
Guesser calls: 440
Completed: Sep 9, 2026, 8:31:53 AM

Reliability

Output contract clean.

All 440 evaluated outputs matched the public structured-action contract.

Provenance

Execution: BX-20260909-B-0003-experimental-M0022-002
Model ID: M-0022
Benchmark: B-0003
Base seed: 0
Git commit: 2d5409df2ba6

Recorded cost

By role.

Cost composition.Share of full-run cost

Guesser78.8%$5.27
Primary Oracle11.2%$0.75
Reviewer5.4%$0.36
Judge4.5%$0.30
Validator0.1%$0.01

Excluded repair overhead$0.171 superseded attempt across 1 affected trial.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-6-astra

Asks the questions and submits the scored guess.

Calls: 440
Cost: $5.2672
Reasoning: High
Routing: Exact provider · openai
Resolved model: openai/gpt-6-astra
Resolved provider: OpenAI
Prompt contract: stateful-category-guesser-v16-five-answer-category-guide
Configuration: M-0022

Resolved provider details

OpenAI440 calls$5.26722,443.18 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: 271
Cost: $0.7492
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

OpenAI271 calls$0.74922,513.18 s

Fallback calls: 0
Provider unreported: 0

Reviewer

google/gemini-3.5-flash-lite

Checks each Oracle YES or NO independently.

Calls: 258
Cost: $0.3588
Reasoning: Medium
Routing: Exact provider · google-ai-studio
Resolved provider: Google AI Studio

Resolved provider details

Google AI Studio258 calls$0.3588447.59 s

Fallback calls: 0
Provider unreported: 0

Judge

anthropic/claude-opus-5

Decides when the Oracle and Reviewer disagree.

Calls: 12
Cost: $0.3026
Reasoning: Medium
Routing: OpenRouter automatic routing
Resolved provider: Claude Platform on AWS

Resolved provider details

Claude Platform on AWS12 calls$0.3026100.37 s

Fallback calls: 0
Provider unreported: 0

Guess Validator

openai/gpt-5.6-luna

Checks a submitted guess against the trusted subject.

Calls: 32
Cost: $0.0062
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

OpenAI32 calls$0.006257.58 s

Fallback calls: 0
Provider unreported: 0