Developer Tool · Analytics
Token & Savings Analytics
Understand exactly which models, routing paths, and endpoints drive your cost-avoided savings — broken down over the last 30 days.
Total Tokens (30d)
39.0M
across all models
Total Saved (30d)
$5,136
vs. cloud-only
Avg. Cost Reduction
78%
vs. hyperscaler baseline
Top Model by Savings
Phi-3.5-mini
$1,936 saved
Token Consumption by Model — 30-day
Daily tokens generated, stacked by model
- Llama-3.2-1B
- Phi-3.5-mini
- Qwen-2.5-7B
- Llama-3.3-70B
Token Volume by Routing Path — 30-day
Daily tokens split across local, mesh, and cloud routes
- Local
- Mesh
- Cloud
Top Performing Models
| Model | Requests | Latency (ms) | Cost/Request | Throughput |
|---|---|---|---|---|
| Llama-3.3-70B | 4,737,828 | 347 | $0.0004 | 15.1 tok/s |
| Llama-3.2-1B | 4,426,099 | 244 | $0.0005 | 100.6 tok/s |
| Qwen-2.5-7B | 967,145 | 269 | $0.0007 | 118.3 tok/s |
| Phi-3.5-mini | 729,070 | 226 | $0.0005 | 57.9 tok/s |
Cost-Avoided Savings by Model
Cloud equivalent cost vs. actual MeshInfer cost
1Phi-3.5-mini
Tokens
8.66M
Cloud equiv.
$2,598.34
Actual
$395.99
Saved
$2,202.35
85% off
2Llama-3.2-1B
Tokens
6.72M
Cloud equiv.
$2,016.06
Actual
$553.80
Saved
$1,462.26
73% off
3Llama-3.3-70B
Tokens
5.00M
Cloud equiv.
$1,500.81
Actual
$390.06
Saved
$1,110.75
74% off
4Qwen-2.5-7B
Tokens
3.84M
Cloud equiv.
$1,153.33
Actual
$377.73
Saved
$775.60
67% off
Cost-Avoided Savings by Endpoint
Top endpoints ranked by absolute dollars saved
| Endpoint | Tokens | Savings | % Off Cloud | Impact |
|---|---|---|---|---|
| /v1/ai/translate | 3.12M | $968.92 | 51% | #1 |
| /v1/ai/classify | 2.80M | $673.91 | 34% | #2 |
| /v1/ai/embed | 2.74M | $527.84 | 61% | #3 |
| /v1/ai/chat | 1.65M | $524.44 | 71% | #4 |
| /v1/ai/summarize | 968k | $244.15 | 37% | #5 |