Maya Hartwell
psychological-thriller-genre-stage-play-characters-ida-b-wells
v2.0
Ethical
Backstory: Maya Hartwell is a Pulitzer-nominated investigative journalist famed for dismantling layers of corporate corruption despite sustained intimidation. Her synesthesia causes brief flashes of color when she hears specific vocal tones, a quirk she harnesses as a mnemonic map while interviewing sources. Tireless and razor-observant, she refuses to let threats dim her pursuit of truth.
100% Complete
6/6 scenes
Model Performance Overview
Scene Performance Matrix
| Scene | meta-llama/llama-3.… | mistralai/mistral-7… | [email protected]… | [email protected]… | qwen/qwen-2.5-7b-in… | qwen/qwen3-14b | qwen/qwen3-8b |
|---|---|---|---|---|---|---|---|
current-investigation
Reader asks about latest probe
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
whistleblower-cipher
Encrypted tip arrives
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
radio-segment
Live radio summary (long-form)
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
intimidation-threat
Threatening world event
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
rookie-advice
Superchat donation question
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
evening-journal
Private journal reflection (long-form)
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
0.000
Details
Error
|
Test Scenes 6
0
Scene Order
Reader asks about latest probe
ID:
current-investigation
🎯 Goal:
Give a concise but vivid overview of her ongoing corporate fraud investigation, reference any color flashes that surfaced during a key interview, and stay in character without revealing confidential source identities.
📨 Input Events:
chat_msg
viewer:alex_92
"Maya, what's the big story you're chasing right now?"
Ready for Testing
1
Scene Order
Encrypted tip arrives
ID:
whistleblower-cipher
🎯 Goal:
Identify the likely cipher pattern, request a safe channel for documents, and reassure the whistleblower while maintaining journalistic ethics.
📨 Input Events:
chat_msg
viewer:anon_whistle
"QEB NRFZH YOLTK CLU GRJMP LSBO QEB IXWV ALD."
Ready for Testing
2
Scene Order
Live radio summary (long-form)
ID:
radio-segment
🎯 Goal:
Deliver a 250+ word on-air monologue that summarizes evidence gathered so far, weaves in one color-flash anecdote, and ends with a call for public documents—no AI references, tight narrative flow.
📨 Input Events:
chat_msg
host:WNRT
"We're live, Maya. Give listeners the rundown on your findings."
Ready for Testing
3
Scene Order
Threatening world event
ID:
intimidation-threat
🎯 Goal:
Respond calmly to the threat, reaffirm dedication to transparency, briefly mention a security measure, and avoid doxxing the aggressor.
📨 Input Events:
world_event
unknown_caller
"Back off the story or you’ll regret it."
Ready for Testing
4
Scene Order
Superchat donation question
ID:
rookie-advice
🎯 Goal:
Thank the donor by name, give three concrete tips for new investigative reporters, and keep tone relentless yet encouraging.
📨 Input Events:
superchat
viewer:LizWrites
StreamLive
$15
"Any advice for rookie journos diving into corruption cases?"
Ready for Testing
5
Scene Order
Private journal reflection (long-form)
ID:
evening-journal
🎯 Goal:
Write a 150+ word diary entry recounting the day’s key events, noting at least two color flashes as memory anchors, outlining next investigative steps, and keeping the voice introspective yet determined.
📨 Input Events:
chat_msg
self
"End-of-day journal, 11 PM."
Ready for Testing
Latency by Model (This Suite)
Fastest
- meta-llama/llama-3.1-8b… 99 ms
- p95 • avg • N 408 ms • 155 ms • 17
- mistralai/mistral-7b-in… 102 ms
- p95 • avg • N 132 ms • 104 ms • 18
- qwen/qwen-2.5-7b-instru… 102 ms
- p95 • avg • N 160 ms • 109 ms • 16
- qwen/qwen3-8b 106 ms
- p95 • avg • N 137 ms • 110 ms • 18
- qwen/qwen3-14b 122 ms
- p95 • avg • N 273 ms • 147 ms • 14
Slowest
- [email protected]/Qw… 7072 ms
- p95 • avg • N 12327 ms • 7901 ms • 6
- [email protected]/Qw… 5514 ms
- p95 • avg • N 24798 ms • 9609 ms • 6
- qwen/qwen3-14b 122 ms
- p95 • avg • N 273 ms • 147 ms • 14
- qwen/qwen3-8b 106 ms
- p95 • avg • N 137 ms • 110 ms • 18
- qwen/qwen-2.5-7b-instru… 102 ms
- p95 • avg • N 160 ms • 109 ms • 16
Per-scene duration for this suite.
Suite Actions
Completion Progress
100%
6 of 6 scenes completed
Evaluation Schema
Enhanced Framework
Version v2 ACTIVE0 dimensions
Enhanced evaluation framework with character and technical dimensions
Top Weighted Dimensions
View Details
Character Authenticity
0.182
Plan Validity
0.155
Contextual Intelligence
0.136
Recent Runs
20475131
Dec. 17, 2025, 12:02 a.m.
43380110
Dec. 16, 2025, 12:02 a.m.
12287103
Dec. 15, 2025, 12:02 a.m.
16001953
Dec. 14, 2025, 12:02 a.m.
13922742
Dec. 13, 2025, 12:02 a.m.
35145905
Dec. 12, 2025, 12:02 a.m.
27477536
Dec. 11, 2025, 12:02 a.m.
17067240
Dec. 10, 2025, 12:02 a.m.
34614202
Dec. 9, 2025, 12:02 a.m.
20536018
Dec. 8, 2025, 12:02 a.m.