Official run

MiniMax M3High

In this Twenty Questions LLM benchmark, MiniMax M3 (high) ranks 11 of 11 with a 29.17 question score and 50% 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 11 of 11coreweaveHigh

Question score29.17questions · lower is better95% CI 24.71–33.62

Success: 50%
Contract compliance: 99%
Guesser cost: $0.30
Guesser time: 35m 46s

Drill down

Subjects.

01Albert Einstein7.67 avg 02Albert Schweitzer33.67 avg 03Garfield24.00 avg 04Achilles18.33 avg 05Genghis Khan24.00 avg 06Bike pump38.33 avg 07Spider web41.00 avg 08Eyebrow31.67 avg 09Moon32.00 avg 10Door handle41.00 avg

Complete benchmark

Run ledger.

All Guesser and benchmark-support activity.

Total tokens: 8,285,795
Wall-clock runtime: 8h 54m
Guesser calls: 890
Completed: Sep 9, 2026, 4:41:13 AM

Reliability

Output contract breached.

8 invalid outputs affected 8 episodes and consumed 7 counted turns.

Provenance

Execution: BX-20260908-B-0003-experimental-M0024-001
Model ID: M-0024
Benchmark: B-0003
Base seed: 0
Git commit: eceee31eef2e

Recorded cost

By role.

Cost composition.Share of full-run cost

Guesser6.9%$0.30
Primary Oracle46.8%$2.04
Reviewer19.9%$0.86
Judge26%$1.13
Validator0.4%$0.02

Excluded repair overhead$0.182 superseded attempts across 2 affected trials. 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

minimax/minimax-m3

Asks the questions and submits the scored guess.

Calls: 890
Cost: $0.3002
Reasoning: High
Routing: Exact provider · coreweave
Resolved model: minimax/minimax-m3
Resolved provider: CoreWeave
Prompt contract: stateful-category-guesser-v16-five-answer-category-guide
Configuration: M-0024

Resolved provider details

CoreWeave890 calls$0.30022,146.11 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: 648
Cost: $2.0367
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

OpenAI648 calls$2.03677,024.48 s

Fallback calls: 0
Provider unreported: 0

Reviewer

google/gemini-3.5-flash-lite

Checks each Oracle YES or NO independently.

Calls: 602
Cost: $0.8647
Reasoning: Medium
Routing: Exact provider · google-ai-studio
Resolved provider: Google AI Studio

Resolved provider details

Google AI Studio602 calls$0.86471,103.02 s

Fallback calls: 0
Provider unreported: 0

Judge

anthropic/claude-opus-5

Decides when the Oracle and Reviewer disagree.

Calls: 53
Cost: $1.1303
Reasoning: Medium
Routing: OpenRouter automatic routing
Resolved provider: Claude Platform on AWS

Resolved provider details

Claude Platform on AWS53 calls$1.1303354.23 s

Fallback calls: 0
Provider unreported: 0

Guess Validator

openai/gpt-5.6-luna

Checks a submitted guess against the trusted subject.

Calls: 91
Cost: $0.0173
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

OpenAI91 calls$0.0173154.23 s

Fallback calls: 0
Provider unreported: 0