Analytics
Token usage, cost, and quotas across all models and workspaces · last 30 days
Requests
128.4K
▲ 12.3% vs prior 30d
Tokens In
42.1M
▲ 8.1% vs prior 30d
Tokens Out
8.3M
▲ 5.4% vs prior 30d
Cost (USD)
$284.50
▲ 4.2% vs prior 30d
Token Usage — Model × Time window
Updated 8s ago
| Model | Minute | Hour | Day | Month | Year |
|---|---|---|---|---|---|
| All models | 12.4K / 7.1K∞ | 680K / 420K∞ | 4.8M / 2.9M∞ | 11.4M / 6.8M27.0M | 94.1M / 56.2M324M |
| claude-opus-4.7 | 3.2K / 1.8K20K | 204K / 128K800K | 1.7M / 1.1M3.5M | 4.2M / 2.6M10M | 36.8M / 22.1M120M |
| claude-sonnet-4.6 | 5.4K / 3.2K15K | 312K / 192K600K | 2.1M / 1.3M2.4M | 4.6M / 2.8M7M | 38.4M / 23.0M84M |
| gpt-5.5 | 2.1K / 1.4K10K | 112K / 71K400K | 812K / 514K1.6M | 1.8M / 1.1M5M | 14.2M / 8.6M60M |
| deepseek-v3 | 1.2K / 0.6K5K | 38K / 22K200K | 142K / 84K800K | 612K / 384K3M | 3.2M / 1.9M36M |
| qwen2.5-72b | 0.5K / 0.1K5K | 14K / 7K200K | 46K / 18K800K | 202K / 98K2M | 1.5M / 0.6M24M |
Each cell shows
read / write | quota. read = prompt tokens, write = completion tokens. quota is the workspace-level total (read + write). ∞ = no quota set at this window. Cells turn amber at 80% and red at 95%. Per-team quotas are edited on Team detail.
| Model | Minute | Hour | Day | Month | Year |
|---|---|---|---|---|---|
| All models | 8.1K / 0.4K | 412K / 24K | 2.8M / 168K | 6.4M / 384K | 52.1M / 3.1M |
| claude-opus-4.7 | 2.4K / 0.1K | 142K / 8K | 1.1M / 62K | 2.6M / 156K | 21.4M / 1.3M |
| claude-sonnet-4.6 | 4.2K / 0.3K | 218K / 14K | 1.4M / 88K | 3.1M / 188K | 25.8M / 1.5M |
| gpt-5.5 | 1.1K / 0K | 42K / 2K | 218K / 14K | 582K / 36K | 4.2M / 240K |
| deepseek-v3 | 0.3K / 0K | 8K / 0K | 42K / 2K | 116K / 4K | 680K / 38K |
| qwen2.5-72b | 0.1K / 0K | 2K / 0K | 14K / 0K | 38K / 0K | 186K / 12K |
cache_read = tokens served from prompt cache (don't bill against quota). cache_write = tokens that filled cache. Cache tokens do not count toward the workspace quota — they're reported here so you can size cache breakpoints and estimate savings. Source: AIGW portkey_user_model_cache_*_tokens_total series.
Tokens — last 30 days
Stacked area placeholder — read (light) + write (dark) over 30d. Wire to
/v1/admin/usage?window=30d&group=model.Quota — this month
Tokens18.2M / 27.0M
Cost (est.)$2,704 / $4,000
claude-sonnet-4.698%
claude-opus-4.768%
Top models (MTD)
1
claude-sonnet-4.6
7.4M tokens
2
claude-opus-4.7
6.8M tokens
3
gpt-5.5
2.9M tokens
4
deepseek-v3
996K tokens
5
qwen2.5-72b
300K tokens
Top teams (MTD)
1
projectx-prod-core
44%
8.1M · 32 users
2
projectx-prod-rag
25%
4.6M · 18 users
3
projectx-prod-eval
17%
3.2M · 12 users
Per-window peaks
Peak TPM 284K · today 14:23
Peak TPH 1.04M · yesterday 11:00
Peak TPD 4.8M · 2026-05-30