Now open — sign in and ship
/v1 endpoint for frontier OSS LLMs.
Kimi · MiMo · DeepSeek · Qwen · GLM behind a single
OpenAI-compatible endpoint at your-org.dynoyard.app/v1.
Drop-in for any OpenAI SDK. Pay-as-you-go from credits.
Built for teams shipping AI products — agent platforms, RAG pipelines, multi-tenant SaaS, LangChain and LiteLLM stacks. Swap one base URL, keep your SDK. Your subdomain, your keys, per-tenant usage.
Python — OpenAI SDK, base_url swap
No SDK changefrom openai import OpenAI
client = OpenAI(
base_url="https://your-org.dynoyard.app/v1",
api_key="sk-dyno-...",
)
resp = client.chat.completions.create(
model="kimi-k2-thinking",
messages=[{"role": "user", "content": "ship it"}],
)We test what's good, ditch what isn't, keep the list tight. Same OpenAI-shaped API for all. Updated as new models ship.
gemini-3-5-flash
1M context · tools · streaming · reasoning · vision · structured_output
Gemini 3.5 Flash — Google's fast multimodal model. 1M-token context, accepts text, image, audio, video, and PDF, with adjustable thinking effort for coding and agentic workloads.
$1.65 in · $9.90 out / 1M tokens
glm-5-2
195K context · tools · streaming · reasoning · structured_output
Zhipu's GLM-5.2 — open flagship for long-horizon coding & agentic tasks, 1M context.
$1.65 in · $4.95 out / 1M tokens
kimi-k2-7
262K context · tools · streaming · reasoning · structured_output
Moonshot's Kimi K2.7 — a large open model strong at agentic coding and tool use, with a 262K-token context.
$0.98 in · $4.08 out / 1M tokens
minimax-m3
1M context · tools · streaming · reasoning · structured_output
M3 reaches frontier-level performance on specialized tasks such as coding and agentic work. It uses MSA (MiniMax Sparse Attention), a new attention architecture proposed by our team, and supports ultra-long context windows of up to 1M tokens. To much anticipation, it is also a natively multimodal model that supports image and video input and can operate a desktop computer.
$0.33 in · $1.32 out / 1M tokens
qwen-plus-3-7
1M context · tools · streaming · reasoning · structured_output
Alibaba's Qwen3.7 Plus — a fast, cost-efficient generalist for high-volume chat and agent workloads, with a 1M-token context.
$0.30 in · $1.21 out / 1M tokens
qwen-plus-3-6
1M context · tools · streaming · reasoning · vision
Alibaba Qwen3.6 Plus — multimodal (image+video in), 1M ctx, function calling, structured output. Non-thinking by default.
$0.55 in · $3.30 out / 1M tokens
qwen-max-3-7
1M context · tools · streaming · reasoning
Alibaba Qwen3.7 Max, 262K ctx, agent-tier.
$0.91 in · $2.72 out / 1M tokens
glm-5-1
203K context · tools · streaming · reasoning
Zhipu / ZAI GLM-5.1 — 200K ctx, strong tool use, top OSS Chinese-English.
$1.08 in · $3.39 out / 1M tokens
deepseek-v4-pro
1M context · tools · streaming · reasoning
DeepSeek V4 Pro — SOTA open-weight reasoning, SWE-Bench 80.6%.
$0.48 in · $0.96 out / 1M tokens
mimo-v2-5-pro
1M context · tools · streaming · reasoning
Xiaomi MiMo V2.5 Pro — 1T MoE / 42B active, 1M ctx, SWE-Bench 78.9%.
$0.48 in · $0.96 out / 1M tokens
OpenRouter is great. We're great too — and built for teams that want isolation, observability, and ownership without standing up infra.
your-org.dynoyard.app/v1 is yours.
Per-org isolation, no shared-tenant noisy-neighbor surprises.
White-label feel without the white-label price.
Issue, rotate, revoke sk-dyno-…
tokens per app. Set monthly spend caps before a runaway agent
eats your weekend.
Per-request token counts, latency, cost, error rate — charted per app. Debug a slow agent without grep'ing your own logs.
Top up via card or wire, spend down per request, auto-refill when balance dips below your threshold. No API subscription, no minimum.
Long agent loops run cheap on cached-prefix pricing. Several catalog models carry 200K+ context windows for long-haul retrieval, codebase work, and multi-turn agents — exact context length is on each model in the catalog above.
Cursor · Zed · OpenCode · Cline · Continue · Aider · LangChain ·
LiteLLM · OpenAI Python · OpenAI JS · curl. Anything that
speaks /v1/chat/completions.
01
Magic-link email. No password. First org auto-created with a slug derived from your email — pick a custom slug later.
02
Add $10 from a card (auto-topup optional). Generate
sk-dyno-… per app, scoped to the models
you want — Cursor key, prod key, dev key.
03
Change one line in your SDK init. Same model names, same response shapes. Watch token counts roll in on the dashboard.
Sign in with your email. Top up $10. Send your first request. Same OpenAI SDK, same prompts, your own subdomain.