NexKit AI · Models
Every id NexKit can route to
One catalogue across both providers. Name any id below in the model field and NexKit sends the request to the provider that owns it — using the credential you attached for that provider.
What resolves today.
Capabilities are what the gateway's dispatch supports for that row — the OpenAI-compatible baseline both providers document for this endpoint. They are not a claim about how well a model does any of them.
| Model id | Provider | Capabilities | Notes |
|---|---|---|---|
| deepseek-v4-flash | DeepSeek | streamingtoolsjson_mode | The DeepSeek default in every config block on this site. |
| deepseek-v4-pro | DeepSeek | streamingtoolsjson_mode | — |
| qwen3.8-max | Qwen | streamingtoolsjson_mode | The flagship, and the Qwen default in every config block on this site. |
| qwen3.7-plus | Qwen | streamingtoolsjson_mode | — |
| qwen3.7-flash | Qwen | streamingtoolsjson_mode | — |
| qwen3-coder-next | Qwen | streamingtoolsjson_mode | The coding-specialised line. Alibaba now points most work at the general-purpose models. |
No context or output limits are published here. Context and output limits are not listed here because NexKit has not measured them. The gateway carries them as unknown rather than as a guess, and each provider publishes its own.
No prices, and no benchmark. You pay your provider directly at their published rates, so their price list is the accurate one; and NexKit has measured no model’s speed or answer quality, so there is nothing here to rank them by.
Ids that no longer work, listed on purpose.
These are the two a person arriving from an older tutorial will try first. The gateway keeps them as retired rows: they never appear in a model list and never resolve, so naming one answers 404 with what is available instead of failing silently.
| Model id | Provider | Status | What happened |
|---|---|---|---|
| deepseek-chat | DeepSeek | Retired | Disabled by DeepSeek on 2026-07-24. Never lists, never resolves. |
| deepseek-reasoner | DeepSeek | Retired | Disabled by DeepSeek on 2026-07-24. Never lists, never resolves. |
Ask your key what it can reach.
This catalogue is the whole gateway. What YOUR key can reach is the subset owned by the providers you have attached — which is what this returns.
curl https://api.nexkit.uk/v1/models -H "Authorization: Bearer nk_live_YOUR_KEY"An empty list is a correct answer. With no provider credential attached, this returns 200 and no models — because a NexKit key on its own reaches nothing. Listing a model that then failed to run would be the worse answer, so the gateway does not.
Which id goes in the config block.
Every config block on this site uses the default below for its provider. Any other id in the table works the same way — nothing else in the configuration changes.
Pick an id and point your agent at it.
One key, one base URL, and the model id is the only thing that changes between providers.