feat(deepinfra): update model YAMLs [bot]

[models-bot]

Auto-generated by poc-agent for provider deepinfra.

---

Note

Low Risk
Metadata-only provider YAML changes (costs, modalities, params, status) with no application logic changes.

Overview
Updates two DeepInfra provider model YAMLs to match current catalog metadata.

FastVideo/LTX-2.3-Distilled-Diffusers now lists audio as an output modality alongside video, consistent with other LTX video models on DeepInfra.

deepseek-ai/DeepSeek-V4-Flash lowers per-token pricing (input, output, and cache-read), adds a reasoning_effort param (none / high / xhigh, default high), and changes status from active to preview.

^{Reviewed by Cursor Bugbot for commit c4eba5e. Bugbot is set up for automated code reviews on this repo. Configure here.}

[github-actions]

/test-models

[harshiv-26]

Gateway test results

• Total: 9
• Passed: 7
• Failed: 1
• Validation failed: 0
• Errored: 0
• Skipped: 1
• Success rate: 87.5%

Provider	Model	Scenarios
deepinfra	FastVideo/LTX-2.3-Distilled-Diffusers	skipped: skip-check
deepinfra	deepseek-ai/DeepSeek-V4-Flash	success: params:stream, params, json-output, tool-call, tool-call:stream, reasoning, reasoning:stream failure: json-output:stream

Failures (1)

deepinfra/deepseek-ai/DeepSeek-V4-Flash — json-output:stream (failure)

Error

Traceback (most recent call last):
  File "/tmp/tmpx4t5337m/snippet.py", line 5, in <module>
    response = client.chat.completions.create(
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/openai/_utils/_utils.py", line 286, in wrapper
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/openai/resources/chat/completions/completions.py", line 1147, in create
    return self._post(
           ^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/openai/_base_client.py", line 1259, in post
    return cast(ResponseT, self.request(cast_to, opts, stream=stream, stream_cls=stream_cls))
                           ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/usr/local/lib/python3.11/site-packages/openai/_base_client.py", line 1047, in request
    raise self._make_status_error_from_response(err.response) from None
openai.AuthenticationError: Error code: 401 - {'status': 'failure', 'message': 'Unauthorized: Forbidden: Invalid token', 'error': {'message': 'Unauthorized: Forbidden: Invalid token', 'type': 'Error', 'code': '401'}, 'error_origin_level': 'authentication'}

Code snippet

from openai import OpenAI

client = OpenAI(api_key="***", base_url="https://internal.devtest.truefoundry.tech/api/llm")

response = client.chat.completions.create(
    model="test-v2-deepinfra/deepseek-ai-DeepSeek-V4-Flash",
    messages=[
        {"role": "user", "content": "List 3 colors with their hex codes in JSON."},
    ],
    response_format={"type": "json_object"},
    stream=True,
)
import json as _json

_accumulated = ""
for chunk in response:
    if chunk.choices and len(chunk.choices) > 0:
        delta = chunk.choices[0].delta
        if delta.content is not None:
            _accumulated += delta.content
            print(delta.content, end="", flush=True)

if not _accumulated:
    raise Exception("VALIDATION FAILED: json-output stream - no content received")

_json.loads(_accumulated)
print("\nVALIDATION: json-output stream SUCCESS")

Successes (7)

deepinfra/deepseek-ai/DeepSeek-V4-Flash — params:stream (success)

Output

The capital of France is Paris.

deepinfra/deepseek-ai/DeepSeek-V4-Flash — params (success)

Output

The capital of France is Paris.

deepinfra/deepseek-ai/DeepSeek-V4-Flash — json-output (success)

Output

{
  "colors": [
    {
      "name": "Crimson Red",
      "hex": "#DC143C"
    },
    {
      "name": "Royal Blue",
      "hex": "#4169E1"
    },
    {
... (truncated, 98 chars omitted)

deepinfra/deepseek-ai/DeepSeek-V4-Flash — tool-call (success)

Output

Function: get_weather
Arguments: {"location": "London"}
VALIDATION: tool-call SUCCESS

deepinfra/deepseek-ai/DeepSeek-V4-Flash — tool-call:stream (success)

Output

{"location": "London"}
VALIDATION: tool-call stream SUCCESS

deepinfra/deepseek-ai/DeepSeek-V4-Flash — reasoning (success)

Output

The expression \(3^{3^{3^{3}}}\) is a power tower, and exponentiation is right-associative, meaning it is evaluated from the top down. So we compute:

... (truncated, 382 chars omitted)

deepinfra/deepseek-ai/DeepSeek-V4-Flash — reasoning:stream (success)

Output

The expression \(3^{3^{3^{3}}}\) is a power tower. Exponentiation is right-associative, meaning we evaluate from the top down. So we compute:

1. Star
... (truncated, 583 chars omitted)

Skipped (1)

deepinfra/FastVideo/LTX-2.3-Distilled-Diffusers — skip-check (skipped)

Skip reason

unsupported mode 'video'