vxcontrol/pentagi
1
0
Fork 0
mirror of https://github.com/vxcontrol/pentagi.git synced 2026-07-22 00:35:27 -04:00
Code Issues 45 Packages Projects Releases 9 Wiki Activity

feat: update model configurations for Anthropic and OpenAI providers

Browse Source 
- Increased max_tokens limits for various models in the Anthropic provider to enhance performance.
- Updated model names and max_tokens in the OpenAI provider, transitioning to the new GPT-5.4 series with adjusted pricing.
- Added new model descriptions for GPT-5.4 series in models.yml to reflect advanced capabilities and pricing structure.
This commit is contained in:
Dmitry Ng
2026-03-19 17:15:31 +03:00
parent bb50d10e04
commit fa70b03ff5
3 changed files with 63 additions and 35 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/pkg/providers/anthropic/config.yml
+11 -11
View File
@@ -2,7 +2,7 @@ simple:
model: claude-haiku-4-5
temperature: 0.5
n: 1
max_tokens: 4000
max_tokens: 8192
price:
input: 1.0
output: 5.0
@@ -13,7 +13,7 @@ simple_json:
model: claude-haiku-4-5
temperature: 0.5
n: 1
max_tokens: 4000
max_tokens: 4096
json: true
price:
input: 1.0
@@ -25,7 +25,7 @@ primary_agent:
model: claude-sonnet-4-5
temperature: 1.0
n: 1
max_tokens: 8192
max_tokens: 16384
reasoning:
max_tokens: 1024
price:
@@ -38,7 +38,7 @@ assistant:
model: claude-sonnet-4-5
temperature: 1.0
n: 1
max_tokens: 8192
max_tokens: 16384
reasoning:
max_tokens: 1024
price:
@@ -51,7 +51,7 @@ generator:
model: claude-opus-4-5
temperature: 1.0
n: 1
max_tokens: 16384
max_tokens: 32768
reasoning:
max_tokens: 4096
price:
@@ -64,7 +64,7 @@ refiner:
model: claude-sonnet-4-5
temperature: 1.0
n: 1
max_tokens: 8192
max_tokens: 20480
reasoning:
max_tokens: 1024
price:
@@ -90,7 +90,7 @@ reflector:
model: claude-haiku-4-5
temperature: 0.7
n: 1
max_tokens: 4000
max_tokens: 4096
reasoning:
max_tokens: 1024
price:
@@ -103,7 +103,7 @@ searcher:
model: claude-haiku-4-5
temperature: 1.0
n: 1
max_tokens: 6000
max_tokens: 8192
reasoning:
max_tokens: 1024
price:
@@ -116,7 +116,7 @@ enricher:
model: claude-haiku-4-5
temperature: 1.0
n: 1
max_tokens: 4000
max_tokens: 4096
price:
input: 1.0
output: 5.0
@@ -127,7 +127,7 @@ coder:
model: claude-sonnet-4-5
temperature: 1.0
n: 1
max_tokens: 16384
max_tokens: 20480
reasoning:
max_tokens: 2048
price:
@@ -140,7 +140,7 @@ installer:
model: claude-sonnet-4-5
temperature: 1.0
n: 1
max_tokens: 8192
max_tokens: 16384
reasoning:
max_tokens: 1024
price:
backend/pkg/providers/openai/config.yml
+24 -24
View File
@@ -1,30 +1,30 @@
simple:
model: gpt-4.1-mini
model: gpt-5.4-nano
temperature: 0.5
top_p: 0.5
n: 1
max_tokens: 4000
max_tokens: 8192
price:
input: 0.4
output: 1.6
cache_read: 0.1
input: 0.2
output: 1.25
cache_read: 0.02
simple_json:
model: gpt-4.1-mini
model: gpt-5.4-nano
temperature: 0.5
top_p: 0.5
n: 1
max_tokens: 4000
max_tokens: 4096
json: true
price:
input: 0.4
output: 1.6
cache_read: 0.1
input: 0.2
output: 1.25
cache_read: 0.02
primary_agent:
model: o4-mini
n: 1
max_tokens: 6000
max_tokens: 16384
reasoning:
effort: medium
price:
@@ -35,7 +35,7 @@ primary_agent:
assistant:
model: o4-mini
n: 1
max_tokens: 6000
max_tokens: 16384
reasoning:
effort: medium
price:
@@ -46,7 +46,7 @@ assistant:
generator:
model: o3
n: 1
max_tokens: 8192
max_tokens: 32768
reasoning:
effort: medium
price:
@@ -57,7 +57,7 @@ generator:
refiner:
model: o3
n: 1
max_tokens: 8192
max_tokens: 20480
reasoning:
effort: high
price:
@@ -66,20 +66,20 @@ refiner:
cache_read: 0.5
adviser:
model: gpt-5.2
model: gpt-5.4
n: 1
max_tokens: 6000
max_tokens: 8192
reasoning:
effort: low
price:
input: 1.75
output: 14.0
cache_read: 0.175
input: 2.5
output: 15.0
cache_read: 0.25
reflector:
model: o4-mini
n: 1
max_tokens: 4000
max_tokens: 4096
reasoning:
effort: medium
price:
@@ -92,7 +92,7 @@ searcher:
temperature: 0.7
top_p: 0.8
n: 1
max_tokens: 4000
max_tokens: 8192
price:
input: 0.4
output: 1.6
@@ -103,7 +103,7 @@ enricher:
temperature: 0.7
top_p: 0.8
n: 1
max_tokens: 4000
max_tokens: 4096
price:
input: 0.4
output: 1.6
@@ -112,7 +112,7 @@ enricher:
coder:
model: o3
n: 1
max_tokens: 16384
max_tokens: 20480
reasoning:
effort: low
price:
@@ -123,7 +123,7 @@ coder:
installer:
model: o4-mini
n: 1
max_tokens: 8192
max_tokens: 16384
reasoning:
effort: low
price:
backend/pkg/providers/openai/models.yml
+28
View File
@@ -1,3 +1,31 @@
# Latest GPT-5.4 series - Frontier models with advanced reasoning and professional workflows
- name: gpt-5.4
description: Best intelligence at scale for agentic, coding, and professional workflows. Flagship model with 1M context window and configurable reasoning effort (none/low/medium/high/xhigh). Excels at complex professional work, sophisticated security research, and advanced autonomous penetration testing requiring maximum cognitive depth and accuracy.
thinking: true
release_date: 2026-01-15
price:
input: 2.5
output: 15.0
cache_read: 0.25
- name: gpt-5.4-mini
description: Strongest mini model yet for coding, computer use, and subagents. Enhanced agentic capabilities with 400K context window and configurable reasoning levels. Ideal for high-volume security workloads, systematic vulnerability analysis, and coordinated multi-tool penetration testing with optimal cost-to-intelligence ratio.
thinking: true
release_date: 2026-01-15
price:
input: 0.75
output: 4.5
cache_read: 0.075
- name: gpt-5.4-nano
description: Cheapest GPT-5.4-class model for simple high-volume tasks like classification, data extraction, ranking, and sub-agents. Optimized for speed and cost with 400K context window. Perfect for rapid reconnaissance, bulk vulnerability scanning, and real-time security monitoring with minimal latency.
thinking: true
release_date: 2026-01-15
price:
input: 0.2
output: 1.25
cache_read: 0.02
# Latest GPT-5.2 series - Enhanced agentic models with improved reasoning and tool integration
- name: gpt-5.2
description: Latest flagship agentic model with enhanced reasoning and tool integration. Excels at autonomous security research, complex exploit chain development, and coordinating multi-tool penetration testing workflows. Optimal for sophisticated threat modeling and adaptive attack strategies.