Capabilities Graph
Built for agents that replay context and decode reality. Agent traffic resends the whole conversation or sensor state on every tool call. Here the repeated part hits cache and bills at the cached input rate.
Catalog order
Ornith 1.5 122B
ornith-1-5-122b
TrendingAgentic · Reasoning · Long Context · Code
$0.20
per 1M input
tokens
tokens
$0.05
per 1M cached
input tokens
input tokens
$0.80
per 1M output
tokens
tokens
94.6%
cache hit, last
24 hours
24 hours
Effective input $0.05 per 1M at that hit rate
210.4
output tok/s, model only
DeepSeek V4.1 Flash 1017
deepseek-v4-1-flash-1017
TrendingUltra-fast · MoE · Coding · Math
$0.15
per 1M input
tokens
tokens
$0.01
per 1M cached
input tokens
input tokens
$0.60
per 1M output
tokens
tokens
29.7%
cache hit, last
24 hours
24 hours
Effective input $0.01 per 1M at that hit rate
320.5
output tok/s, model only
GLM 5.3
glm-5-3
FlagshipReasoning · High-Precision Tool Calling · Enterprise
$1.30
per 1M input
tokens
tokens
$0.13
per 1M cached
input tokens
input tokens
$4.20
per 1M output
tokens
tokens
28.1%
cache hit, last
24 hours
24 hours
Effective input $0.13 per 1M at that hit rate
185.0
output tok/s, model only
Ornith 1.5 35B
ornith-1-5-35b
EfficientFast Reasoning · Low Latency · Agentic Workflows
$0.10
per 1M input
tokens
tokens
$0.01
per 1M cached
input tokens
input tokens
$0.40
per 1M output
tokens
tokens
96.8%
cache hit, last
24 hours
24 hours
Effective input $0.01 per 1M at that hit rate
290.0
output tok/s, model only
DeepSeek V4.1 Flash
deepseek-v4-1-flash
High SpeedFast Coding · Low Cost · MoE Architecture
$0.14
per 1M input
tokens
tokens
$0.03
per 1M cached
input tokens
input tokens
$0.58
per 1M output
tokens
tokens
29.2%
cache hit, last
24 hours
24 hours
Effective input $0.03 per 1M at that hit rate
340.0
output tok/s, model only
GLM 5.3 Flash
glm-5-3-flash
TrendingReasoning · Tool Use · Multimodal · Long Context
$0.10
per 1M input
tokens
tokens
$0.01
per 1M cached
input tokens
input tokens
$0.40
per 1M output
tokens
tokens
98.9%
cache hit, last
24 hours
24 hours
Effective input $0.011 per 1M at that hit rate
254.1
output tok/s, model only
Nemotron 3.5 Lightning 30B
nemotron-3-5-lightning-30b
Reasoning · Tool Use · Code
$0.05
per 1M input
tokens
tokens
$0.01
per 1M cached
input tokens
input tokens
$0.15
per 1M output
tokens
tokens
87.7%
cache hit, last
24 hours
24 hours
Effective input $0.015 per 1M at that hit rate
540.9
output tok/s, model only
Qwen3.8 27B
qwen3-8-27b
Reasoning · Tool Use · Multimodal
$0.10
per 1M input
tokens
tokens
$0.01
per 1M cached
input tokens
input tokens
$0.40
per 1M output
tokens
tokens
96.1%
cache hit, last
24 hours
24 hours
Effective input $0.014 per 1M at that hit rate
140
output tok/s, end to end
DeepSeek-V3
deepseek-v3
Reasoning · Multimodal · Tool Use
$0.10
per 1M input
tokens
tokens
$0.01
per 1M cached
input tokens
input tokens
$0.40
per 1M output
tokens
tokens
97.6%
cache hit, last
24 hours
24 hours
Effective input $0.013 per 1M at that hit rate
216
output tok/s, model only
Physical Intelligence pi0
pi0-robotics
Physical Control · Spatial Reasoning · Proprioception
$0.15
per 1M input
tokens
tokens
$0.01
per 1M cached
input tokens
input tokens
$0.50
per 1M output
tokens
tokens
88.5%
cache hit, last
24 hours
24 hours
Effective physical execution $0.02 per 1M at that hit rate
85
hz control loop, edge
Nvidia Cosmos 1.0
nvidia-cosmos
World Modeling · Simulation · Video Understanding
$0.25
per 1M input
tokens
tokens
$0.02
per 1M cached
input tokens
input tokens
$0.80
per 1M output
tokens
tokens
84.1%
cache hit, last
24 hours
24 hours
Effective inference $0.04 per 1M at that hit rate
45
fps simulation, remote
Llama 3.1 405B
llama-3-1-405b
Reasoning · Tool Use · Planning
$0.80
per 1M input
tokens
tokens
$0.08
per 1M cached
input tokens
input tokens
$2.00
per 1M output
tokens
tokens
91.2%
cache hit, last
24 hours
24 hours
Effective input $0.12 per 1M at that hit rate
120
output tok/s, model only
Mistral Large 2
mistral-large-2
Language · Reasoning · Tool Use
$2.00
per 1M input
tokens
tokens
$0.20
per 1M cached
input tokens
input tokens
$6.00
per 1M output
tokens
tokens
82.4%
cache hit, last
24 hours
24 hours
Effective input $0.53 per 1M at that hit rate
60
output tok/s, model only
Llama 3.2 90B Vision
llama-3-2-90b-vision-instruct
Vision · Perception · Multimodal
$0.80
per 1M input
tokens
tokens
$0.08
per 1M cached
input tokens
input tokens
$2.00
per 1M output
tokens
tokens
94.2%
cache hit, last
24 hours
24 hours
Effective input $0.12 per 1M at that hit rate
115
output tok/s, model only
TimesFM 1.0
timesfm-1-0-200m
Temporal Reasoning · Prediction
$0.01
per 1M input
tokens
tokens
$0.001
per 1M cached
input tokens
input tokens
$0.05
per 1M output
tokens
tokens
75.0%
cache hit, last
24 hours
24 hours
Effective input $0.008 per 1M at that hit rate
3200
series steps/s
Qwen 2.5 Coder 32B
qwen2-5-coder-32b
Code · Verification · Reasoning
$0.10
per 1M input
tokens
tokens
$0.01
per 1M cached
input tokens
input tokens
$0.40
per 1M output
tokens
tokens
98.1%
cache hit, last
24 hours
24 hours
Effective input $0.012 per 1M at that hit rate
185
output tok/s, model only
Command R+
c4ai-command-r-plus
Language · Tool Use · Planning
$3.00
per 1M input
tokens
tokens
$0.30
per 1M cached
input tokens
input tokens
$15.00
per 1M output
tokens
tokens
85.2%
cache hit, last
24 hours
24 hours
Effective input $0.70 per 1M at that hit rate
45
output tok/s, model only