Parasail
Parasail runs open models from many vendors behind one API. In LM15 its
provider is parasail.
| Provider | parasail (litellm spelling: parasail/) |
| Key | PARASAIL_API_KEY, from saas.parasail.io/keys |
| Model names | the vendor’s name: parasail:meta-llama/Llama-3.3-70B-Instruct |
| Needs | Python 1.1.0, TypeScript 1.0.0-rc.2, Rust 1.0.0-rc.2, Go v1.1.0-rc.2 or later |
Chat, streaming, tools (including a forced tool choice), structured output and the model list work as with any provider. Measured against Parasail on September 26, 2026:
- Reasoning off is refused by Parasail, with a clear error, on gpt-oss,
which can’t stop reasoning. Some models (DeepSeek V3.1, Qwen3.5) turn
thinking on and off through their own settings; pass those in the
request’s
extensions. - Images inside a tool result reach the model.
- A model that doesn’t exist raises
UnsupportedModelError, as on every provider, though Parasail words it as a missing deployment.