Configuration
Models
gemma-4-local local FREE
provider:
local (llama.cpp)
model_id:
gemma-4-E2B-it-UD-Q6_K_XL
api_key_env:
none (no auth required)
api_base:
redacted - local installations configure their own endpoint
context:
32,768
timeout:
120s
max out:
16,384
/32,768
cost in:
0
cost out:
0
per 1K tokens (local GPU)
Devil's Advocate can run entirely on open-weight models. Commercial API access not required. Any OpenAI-compatible endpoint works. Assign local models to every review role for a self-hosted, enterprise-controlled deployment.
A low parameter, quantized model will produce noticeably weaker adversarial review than state of the art commercial models. This local model is provided to demonstrate flexibility.
claude-haiku-4-5 $$$
provider:
anthropic
model_id:
claude-haiku-4-5-20251001
api_key_env:
ANTHROPIC_API_KEY
context:
200,000
timeout:
120s
max out:
10000
/64,000
cost in:
0.0008
cost out:
0.004
per 1K tokens
claude-opus-4-5 $$$$$
provider:
anthropic
model_id:
claude-opus-4-5
api_key_env:
ANTHROPIC_API_KEY
context:
200,000
timeout:
900s
max out:
19200
/64,000
cost in:
0.005
cost out:
0.025
per 1K tokens
claude-opus-4-6 $$$$$
provider:
anthropic
model_id:
claude-opus-4-6
api_key_env:
ANTHROPIC_API_KEY
context:
200,000
timeout:
900s
max out:
38400
/128,000
cost in:
0.005
cost out:
0.025
per 1K tokens
claude-sonnet-4-5 $$$$$
provider:
anthropic
model_id:
claude-sonnet-4-5-20250929
api_key_env:
ANTHROPIC_API_KEY
context:
200,000
timeout:
900s
max out:
19200
/64,000
cost in:
0.003
cost out:
0.015
per 1K tokens
claude-sonnet-4-6 $$$$$
provider:
anthropic
model_id:
claude-sonnet-4-6
api_key_env:
ANTHROPIC_API_KEY
context:
200,000
timeout:
900s
max out:
19200
/64,000
cost in:
0.003
cost out:
0.015
per 1K tokens
deepseek-v4-flash $
provider:
openai
model_id:
deepseek-v4-flash
api_key_env:
DEEPSEEK_API_KEY
api_base:
https://api.deepseek.com
context:
1,000,000
timeout:
900s
max out:
16384
/384,000
cost in:
0.00014
cost out:
0.00028
per 1K tokens
deepseek-v4-pro $
provider:
openai
model_id:
deepseek-v4-pro
api_key_env:
DEEPSEEK_API_KEY
api_base:
https://api.deepseek.com
context:
1,000,000
timeout:
900s
max out:
16384
/384,000
cost in:
0.000435
cost out:
0.00087
per 1K tokens
gemini-2.5-pro $$$$
provider:
openai
model_id:
gemini-2.5-pro
api_key_env:
GOOGLE_API_KEY
api_base:
https://generativelanguage.googleapis.com/v1beta/openai
context:
1,048,576
timeout:
900s
max out:
19660
/65,536
cost in:
0.00125
cost out:
0.01
per 1K tokens
gemini-3-flash-preview $$
provider:
openai
model_id:
gemini-3-flash-preview
api_key_env:
GOOGLE_API_KEY
api_base:
https://generativelanguage.googleapis.com/v1beta/openai
context:
1,048,576
timeout:
900s
max out:
19660
/65,535
cost in:
0.0005
cost out:
0.003
per 1K tokens
gemini-3-pro-preview $$$$
provider:
openai
model_id:
gemini-3-pro-preview
api_key_env:
GOOGLE_API_KEY
api_base:
https://generativelanguage.googleapis.com/v1beta/openai
context:
1,048,576
timeout:
900s
max out:
19660
/65,536
cost in:
0.002
cost out:
0.012
per 1K tokens
gemini-3.1-pro-preview $$$$
provider:
openai
model_id:
gemini-3.1-pro-preview
api_key_env:
GOOGLE_API_KEY
api_base:
https://generativelanguage.googleapis.com/v1beta/openai
context:
1,048,576
timeout:
900s
max out:
19660
/65,536
cost in:
0.002
cost out:
0.012
per 1K tokens
gpt-5.1 $$$$
provider:
openai
model_id:
gpt-5.1
api_key_env:
OPENAI_API_KEY
api_base:
https://api.openai.com/v1
context:
272,000
timeout:
120s
max out:
38400
/128,000
cost in:
0.00125
cost out:
0.01
per 1K tokens
use_completion_tokens: true
gpt-5.2 $$$$$
provider:
openai
model_id:
gpt-5.2
api_key_env:
OPENAI_API_KEY
api_base:
https://api.openai.com/v1
context:
128,000
timeout:
900s
max out:
15000
/128,000
cost in:
0.005
cost out:
0.015
per 1K tokens
use_completion_tokens: true
gpt-5.2-codex $$$$
provider:
openai
model_id:
gpt-5.2-codex
api_key_env:
OPENAI_API_KEY
api_base:
https://api.openai.com/v1
context:
400,000
timeout:
180s
max out:
38400
/128,000
cost in:
0.00175
cost out:
0.014
per 1K tokens
use_completion_tokens: true
gpt-5.2-pro $$$$
provider:
openai
model_id:
gpt-5.2-pro
api_key_env:
OPENAI_API_KEY
api_base:
https://api.openai.com/v1
context:
400,000
timeout:
900s
max out:
38400
/128,000
cost in:
0.00175
cost out:
0.014
per 1K tokens
use_completion_tokens: true
gpt-5.3-codex $$$$
provider:
openai
model_id:
gpt-5.3-codex
api_key_env:
OPENAI_API_KEY
api_base:
https://api.openai.com/v1
context:
400,000
timeout:
180s
max out:
38400
/128,000
cost in:
0.00175
cost out:
0.014
per 1K tokens
use_completion_tokens: true
gpt-5.3-codex-spark disabled $$$$
provider:
openai
model_id:
gpt-5.3-codex-spark
api_key_env:
OPENAI_API_KEY
api_base:
https://api.openai.com/v1
context:
128,000
timeout:
60s
max out:
38400
/128,000
cost in:
0.00175
cost out:
0.014
per 1K tokens
use_completion_tokens: true
grok-4-0709 $$$$$
provider:
openai
model_id:
grok-4-0709
api_key_env:
XAI_API_KEY
api_base:
https://api.x.ai/v1
context:
256,000
timeout:
900s
max out:
39321
/131,072
cost in:
0.003
cost out:
0.015
per 1K tokens
grok-4-1-fast-reasoning $$$$$
provider:
openai
model_id:
grok-4-1-fast-reasoning
api_key_env:
XAI_API_KEY
api_base:
https://api.x.ai/v1
context:
2,000,000
timeout:
900s
max out:
15000
/30,000
cost in:
0.003
cost out:
0.015
per 1K tokens
grok-code-fast-1 $
provider:
openai
model_id:
grok-code-fast-1
api_key_env:
XAI_API_KEY
api_base:
https://api.x.ai/v1
context:
256,000
timeout:
900s
max out:
3000
/10,000
cost in:
0.0002
cost out:
0.0015
per 1K tokens
kimi-k2-thinking $$
provider:
openai
model_id:
kimi-k2-thinking
api_key_env:
MOONSHOT_API_KEY
api_base:
https://api.moonshot.ai/v1
context:
262,144
timeout:
900s
max out:
19660
/65,535
cost in:
0.0006
cost out:
0.0025
per 1K tokens
kimi-k2-thinking-turbo $$$
provider:
openai
model_id:
kimi-k2-thinking-turbo
api_key_env:
MOONSHOT_API_KEY
api_base:
https://api.moonshot.ai/v1
context:
262,144
timeout:
900s
max out:
19660
/65,535
cost in:
0.00115
cost out:
0.008
per 1K tokens
kimi-k2.5 $$$
provider:
openai
model_id:
kimi-k2.5
api_key_env:
MOONSHOT_API_KEY
api_base:
https://api.moonshot.ai/v1
context:
262,144
timeout:
900s
max out:
15000
/65,535
cost in:
0.0006
cost out:
0.003
per 1K tokens
minimax-m2.5 $
provider:
minimax
model_id:
MiniMax-M2.5
api_key_env:
MINIMAX_API_KEY
api_base:
https://api.minimax.io
context:
200,000
timeout:
900s
max out:
19660
/65,536
cost in:
0.0003
cost out:
0.0012
per 1K tokens
minimax-m2.5-fast $$
provider:
minimax
model_id:
MiniMax-M2.5-highspeed
api_key_env:
MINIMAX_API_KEY
api_base:
https://api.minimax.io
context:
200,000
timeout:
300s
max out:
19660
/65,536
cost in:
0.0006
cost out:
0.0024
per 1K tokens
API Keys
Enter demo to use pre-selected models at no cost, or paste your own API keys for full control.
Keys are stored in your session and used only for the current visit.
Loading...
Validation
Configuration is valid.
Role Assignments (demo)
Author:
deepseek-v4-flash
Reviewer 1:
deepseek-v4-flash
Reviewer 2:
minimax-m2.5
Dedup:
minimax-m2.5
Normalization:
deepseek-v4-flash
Revision:
deepseek-v4-flash
Integration:
deepseek-v4-flash
What do these roles do?
Author: Generates initial responses to reviewer findings and produces revised artifacts
Reviewer (x2): Independently analyzes input for issues, risks, and improvements
Dedup: Consolidates overlapping findings from multiple reviewers into groups
Normalization: Standardizes severity levels and categories across grouped findings
Revision: Creates the final revised artifact incorporating accepted feedback
Integration: Reviews cross-system integration concerns and component interactions
Development
live e2e tests:
disabled
run without -m live flag