Model usage and scheduled jobs, with 7-day and 30-day views.
Generated 2026-05-20
Figures and costs on this page are best-effort estimates from the machine that built it.
Per-day session token totals (agent JSONL) and cron job run token totals (runs store), aligned on UTC calendar days. Chart is inline SVG (no network); Y-axis is linear from 0 to the largest single-day total (model or cron). The table has exact counts.
Generated 2026-05-20 ·
Source: [redacted path] ·
Non-xAI token columns = session logs in the window (parseable timestamp). xAI: tokens + $/M from invoice preview (billing cycle); cost from POST …/usage. Configure credentials (local): [REDACTED_AUTH] / XAI_TEAM_ID.
| Model | Provider | $/M In | $/M Out | Requests | Prompt Tokens | Compl. Tokens | Total Tokens | Est. Cost |
|---|---|---|---|---|---|---|---|---|
| deepseek/deepseek-chat deepseek-chat (V3.2). Cache hit: $0.07/M. Max 8K output. | DeepSeek | $0.27 | $1.10 | 207 | 6,220,799 | 55,251 | 6,276,050 | $0.3692 |
| deepseek/deepseek-reasoner deepseek-reasoner (R1). Thinking tokens count toward output cost. | DeepSeek | $0.55 | $2.19 | 10 | 388,714 | 5,808 | 394,522 | $0.0267 |
| google/gemini-2.5-flash Standard context (<200k). Thinking-mode output is $3.50/M. | Google AI Studio | — | — | 98 | 655,287 | 3,855 | 550,139 | $0.0000 |
| xai/grok-4-1-fast-reasoning Grok 4.1 Fast (reasoning + non-reasoning). 2M token context. | xAI | $0.20 | $0.50 | 19 | 0 | 0 | 0 | $0.0000 |
| nvidia/minimax-m2.7 MiniMax official pay-as-you-go pricing for M2.7 standard. | MiniMax via NVIDIA NIM | — | — | 16 | 0 | 0 | 0 | $0.0000 |
| nvidia/deepseek-v4-pro DeepSeek V4-Pro: 1.6T params, 49B active. NVIDIA NIM endpoint. | NVIDIA NIM (DeepSeek V4 Pro) | — | — | 0 | 0 | 0 | 0 | $0.0000 |
| nvidia/deepseek-v4-flash DeepSeek published pricing for V4-Flash. NVIDIA NIM endpoint. | NVIDIA NIM (DeepSeek V4 Flash) | — | — | 16 | 0 | 0 | 0 | $0.0000 |
| groq-main/llama-3.3-70b-versatile Groq LPU inference. 128K context, up to 33K output. | Groq | — | — | 40 | 0 | 0 | 0 | $0.0000 |
| cerebras-8b/llama-3.1-8b Cerebras Llama 3.1 8B. Very fast inference on Cerebras hardware. | Cerebras | — | — | 0 | 0 | 0 | 0 | $0.0000 |
| TOTAL | 406 | 7,264,800 | 64,914 | 7,220,711 | $0.3959 | |||
| Model | Provider | $/M In | $/M Out | Requests | Prompt Tokens | Compl. Tokens | Total Tokens | Est. Cost |
|---|---|---|---|---|---|---|---|---|
| deepseek/deepseek-chat deepseek-chat (V3.2). Cache hit: $0.07/M. Max 8K output. | DeepSeek | $0.27 | $1.10 | 896 | 27,610,348 | 226,517 | 27,836,865 | $1.3779 |
| deepseek/deepseek-reasoner deepseek-reasoner (R1). Thinking tokens count toward output cost. | DeepSeek | $0.55 | $2.19 | 48 | 2,044,188 | 23,352 | 2,067,540 | $0.1217 |
| google/gemini-2.5-flash Standard context (<200k). Thinking-mode output is $3.50/M. | Google AI Studio | — | — | 326 | 2,245,806 | 12,777 | 1,780,567 | $0.0000 |
| xai/grok-4-1-fast-reasoning Grok 4.1 Fast (reasoning + non-reasoning). 2M token context. | xAI | $0.20 | $0.50 | 140 | 0 | 0 | 0 | $0.0000 |
| nvidia/minimax-m2.7 MiniMax official pay-as-you-go pricing for M2.7 standard. | MiniMax via NVIDIA NIM | — | — | 49 | 0 | 0 | 0 | $0.0000 |
| nvidia/deepseek-v4-pro DeepSeek V4-Pro: 1.6T params, 49B active. NVIDIA NIM endpoint. | NVIDIA NIM (DeepSeek V4 Pro) | — | — | 0 | 0 | 0 | 0 | $0.0000 |
| nvidia/deepseek-v4-flash DeepSeek published pricing for V4-Flash. NVIDIA NIM endpoint. | NVIDIA NIM (DeepSeek V4 Flash) | — | — | 50 | 0 | 0 | 0 | $0.0000 |
| groq-main/llama-3.3-70b-versatile Groq LPU inference. 128K context, up to 33K output. | Groq | — | — | 120 | 0 | 0 | 0 | $0.0000 |
| cerebras-8b/llama-3.1-8b Cerebras Llama 3.1 8B. Very fast inference on Cerebras hardware. | Cerebras | — | — | 18 | 0 | 0 | 0 | $0.0000 |
| TOTAL | 1,647 | 31,900,342 | 262,646 | 31,684,972 | $1.4996 | |||
|