Dolphin LLM vs hosted Qwen3.6-35B-A3B on OpenRouter
Updated · prices checked on that date
Dolphin LLM runs a model of the Qwen3.6-35B-A3B class. The same base model is offered by several providers through OpenRouter and directly. Here is how the prices and terms compare.
Key points
- Most providers charge $0.10 per 1M input and $0.89–$1.00 per 1M output tokens; Dolphin LLM charges $0.15 and $0.50.
- OpenRouter adds a 5.5% fee when you buy credit on its Standard plan; Dolphin LLM adds no fee.
- The providers offer a longer context (about 256,000–262,000 tokens) than Dolphin LLM (64,000).
- Output-heavy workloads cost less on Dolphin LLM; input-heavy ones with long prompts can cost less elsewhere.
Prices for the same model class
| Where | Input / 1M | Output / 1M | Context |
|---|---|---|---|
| Dolphin LLM | $0.15 | $0.50 | 64,000 |
| OpenRouter · CoreWeave (fp8) | $0.10 | $0.89 | 262,000 |
| OpenRouter · AkashML, Parasail (fp8) | $0.10 | $0.90 | 262,000 |
| OpenRouter · DeepInfra (fp8) | $0.10 | $0.95 | 262,000 |
| OpenRouter · Venice | $0.10 | $1.00 | 256,000 |
| OpenRouter · Darkbloom (fp4) | $0.05 | $0.70 | 262,000 |
| DeepInfra direct | $0.10 | $0.95 | 256,000 |
OpenRouter passes the provider's list price through without a markup on inference, and charges a 5.5% fee on credit purchases on its Standard plan.
What a request costs
| Request | Dolphin LLM | OpenRouter · CoreWeave, incl. 5.5% fee |
|---|---|---|
| Chat turn: 4,000 in, 2,000 out | $0.0016 | $0.0023 |
| Long answer: 1,000 in, 8,000 out | $0.00415 | $0.0076 |
| Long document: 50,000 in, 1,000 out | $0.0080 | $0.0062 |
The lower output price wins when the model writes a lot: chat, stories, code. When you send long prompts and get short answers, the lower input price of the other providers wins.
Context
The providers serve the model with a context of about 256,000–262,000 tokens. Dolphin LLM serves 64,000 tokens, prompt and answer together. If your requests are longer than that, use a provider with the longer window.
Behaviour and billing
Dolphin LLM serves an uncensored build of the model, tuned to answer lawful adult requests without refusals. Billing is prepaid by card or crypto, without a purchase fee; credit never expires and failed requests are free. OpenRouter routes between providers and gives one bill for many models.
Which one to choose
- Choose OpenRouter if you want many models behind one key or prompts longer than 64,000 tokens.
- Choose Dolphin LLM if your workload is output-heavy, if you want the uncensored build, or if you want prepaid billing without a fee.
Sources
Prices change. Check each provider's page before you decide.
Questions and answers
Is it the same model?
It is the same model class (Qwen3.6-35B-A3B). Dolphin LLM serves an uncensored build of it, so answers can differ from the standard build other providers serve.
Why is the output price different?
Dolphin LLM runs the model on its own GPU servers and sets its own price list: $0.15 per 1M input and $0.50 per 1M output tokens. Providers on OpenRouter set theirs independently.