Providers¶
textfsm-ai integrates 18 LLM providers behind one interface (generate(),
run_pipeline(), and the textfsm-ai generate/textfsm-ai pipeline CLI
commands all take the same provider="..." value). Most providers just
need an API key; a few cloud-gateway providers (Bedrock, Vertex AI, OCI)
use their own ambient credential chains instead and take extra parameters
in place of api_key.
The actual per-provider SDK calls are delegated to
anyask, a standalone
package (same maintainer) that owns the 18 provider implementations —
textfsm-ai's pip install textfsm-ai[<provider>] extras are a thin
pass-through to the matching anyask[<provider>] extra.
Run textfsm-ai providers list at any time for the live, code-derived
version of this table.
All providers¶
| Provider | provider value |
Credential | Extra params | Example model |
|---|---|---|---|---|
| OpenAI | "openai" |
OPENAI_API_KEY |
— | gpt-4o-mini |
| Anthropic | "anthropic" |
ANTHROPIC_API_KEY |
— | claude-haiku-4-5-20251001 |
| Google Gemini | "gemini" |
GEMINI_API_KEY |
— | gemini-2.5-flash |
| DeepSeek | "deepseek" |
DEEPSEEK_API_KEY |
— | deepseek-v4-flash |
| Azure OpenAI | "azure" |
AZURE_API_KEY |
endpoint (AZURE_ENDPOINT), api_version (AZURE_API_VERSION), model is a deployment name (AZURE_DEPLOYMENT) |
(your deployment name) |
| Groq | "groq" |
GROQ_API_KEY |
— | llama-3.1-8b-instant |
| xAI (Grok) | "xai" |
XAI_API_KEY |
— | grok-3-mini |
| Together AI | "together" |
TOGETHER_API_KEY |
— | meta-llama/Llama-3.1-8B-Instruct-Turbo |
| Fireworks AI | "fireworks" |
FIREWORKS_API_KEY |
— | accounts/fireworks/models/llama-v3p1-8b-instruct |
| Cerebras | "cerebras" |
CEREBRAS_API_KEY |
— | llama3.1-8b |
| Perplexity | "perplexity" |
PERPLEXITY_API_KEY |
— | sonar |
| OpenRouter | "openrouter" |
OPENROUTER_API_KEY |
— | google/gemini-2.5-flash-lite |
| Moonshot AI (Kimi) | "moonshot" |
MOONSHOT_API_KEY |
— | moonshot-v1-8k |
| Mistral AI | "mistral" |
MISTRAL_API_KEY |
— | mistral-small-latest |
| Amazon Bedrock | "bedrock" |
(none — AWS credential chain) | region (BEDROCK_REGION/BEDROCK_DEFAULT_REGION, required) |
anthropic.claude-haiku-4-5-v1:0 |
| Cohere | "cohere" |
COHERE_API_KEY |
— | command-light |
| Google Vertex AI | "vertexai" |
(none — GCP ADC credential chain) | project (VERTEXAI_PROJECT, required), region (VERTEXAI_REGION, required) |
gemini-2.5-flash |
| Oracle OCI | "oci" |
(none — ~/.oci/config credential file) |
compartment_id (OCI_COMPARTMENT_ID, required), region (OCI_REGION, optional — falls back to the config file) |
meta.llama-3.3-70b-instruct |
Credentials resolve in this order everywhere: CLI flag / explicit function
argument > provider-specific environment variable > providers.yaml. See
the CLI Guide for the full flag reference and the
Quickstart for Python API usage
examples, including the three cloud-gateway providers above.
Two implementation shapes¶
Internally (inside anyask, not this repo), providers fall into two
groups — this only matters if you're extending provider support itself,
not for calling textfsm-ai:
- Shape A — OpenAI-compatible chat-completions API (DeepSeek, Groq, xAI, Together AI, Fireworks AI, Cerebras, Perplexity, OpenRouter, Moonshot). These share one HTTP client implementation and differ only in base URL, env var, and default model.
- Shape B — a native SDK with its own request/response shape (OpenAI,
Anthropic, Gemini, Azure, Mistral, Bedrock, Cohere, Vertex AI, OCI).
Bedrock, Vertex AI, and OCI additionally take extra constructor
parameters (
region/project/compartment_id) instead ofapi_key, since they authenticate via their cloud platform's own credential chain rather than a project-level API key.
There is no automatic provider routing or fallback — --provider (or
provider="..." in the Python API) is always explicit.