Per-token rates, billed per request. No API subscription, no minimum. Plus what raw model access doesn't give you: a per-org subdomain, native usage dashboard, per-key access + caps, auto-topup, and inline per-request cost on every response.
Per 1M tokens · input / output · USD
| Model | Context | Input | Cached | Output |
|---|---|---|---|---|
| Gemini 3.5 Flash gemini-3-5-flash | 1M | $1.65 | $0.60 | $9.90 |
| GLM-5.2 glm-5-2 | 195K | $1.65 | $0.60 | $4.95 |
| kimi-k2-7 kimi-k2-7 | 262K | $0.98 | $0.35 | $4.08 |
| MiniMax M3 minimax-m3 | 1M | $0.33 | $0.18 | $1.32 |
| Qwen3.7 Plus qwen-plus-3-7 | 1M | $0.30 | $0.12 | $1.21 |
| Qwen3.6 Plus qwen-plus-3-6 | 1M | $0.55 | $0.22 | $3.30 |
| Qwen3.7 Max qwen-max-3-7 | 1M | $0.91 | $0.40 | $2.72 |
| GLM-5.1 glm-5-1 | 203K | $1.08 | $0.30 | $3.39 |
| DeepSeek V4 Pro deepseek-v4-pro | 1M | $0.48 | $0.10 | $0.96 |
| MiMo V2.5 Pro mimo-v2-5-pro | 1M | $0.48 | $0.10 | $0.96 |
Per-million-token rates, billed per request from prepaid credits. Rates subject to change with 30 days' notice.
Worked example: one Cursor-style agent day at 500K input + 50K output tokens. Most teams land between $0.50 and $5 / day per active user.
gemini-3-5-flash
$1.32
per agent-day
glm-5-2
$1.07
per agent-day
kimi-k2-7
$0.69
per agent-day
minimax-m3
$0.23
per agent-day
qwen-plus-3-7
$0.21
per agent-day
qwen-plus-3-6
$0.44
per agent-day
qwen-max-3-7
$0.59
per agent-day
glm-5-1
$0.71
per agent-day
deepseek-v4-pro
$0.29
per agent-day
mimo-v2-5-pro
$0.29
per agent-day
Top up your balance via card from $10. Each successful API call debits cents from credits — the same per-million rate you see above. Balance shown on your dashboard; receipts auto-emailed.
Set a low-balance threshold; we charge your saved card off-session when you dip below. Daily caps + per-app caps keep refills under control.
Larger teams: wire USD direct to our company account. We apply credits with the wire reference, audit-logged. Email hello@dynoyard.app.
Teams of up to 3 free. $10/seat/mo only kicks in beyond that, billed against your credits — no separate invoice. Cancel any time by removing seats.
You can. We're built for teams that want a dedicated subdomain, isolated per-org state, per-key access controls with caps, inline per-request cost, and a native usage dashboard — without setting up infra. Same OpenAI-compatible API, same frontier OSS models, one base-URL swap.
API returns 402 insufficient_credits. Auto-topup
(if enabled) charges your saved card; otherwise top up manually from
the dashboard. Auto-topup has a $1 grace window so an in-flight burst
doesn't 402 mid-request.
Unused credits refundable for 30 days from top-up. Email hello@dynoyard.app. Stripe processes the refund back to the original card; we debit your balance by the matching amount.
Yes — per-request token counts, latency, status, model, key. Last 90 days on dashboard; longer retention available on request. We don't retain prompt or completion bodies for training or analytics — the only exceptions are an opt-in response cache (per-org, short TTL) and a 1KB error-body snippet captured for debugging on failed requests.
Above $5k/mo we'll quote custom rates + dedicated capacity (no shared queue). Reach out: hello@dynoyard.app.