From 79e2537cbb811773d73b17b6a89963ac5a99754b Mon Sep 17 00:00:00 2001 From: jaberjaber23 Date: Tue, 18 Aug 2026 19:00:12 +0300 Subject: [PATCH] feat: add DeepSeek V4 Pro to the runinfra provider Fifth hosted model, USD per token from the published per-1M rates (0.60 input, 1.90 output, 0.03 cached input read), 1,048,576 context, 32,768 output cap. Verified live on the public endpoint on 2026-08-18 before this commit: json_object and strict json_schema 3 of 3 each, one tool call round-tripped, a prefix-cache hit on a repeated identical prompt, reasoning content present on the wire when enabled (reasoning is toggleable on this host), and the output cap refusing over-cap requests with 400. --- ...odel_prices_and_context_window_backup.json | 20 +++++++++++++++++++ model_prices_and_context_window.json | 20 +++++++++++++++++++ .../llms/openai_like/test_json_providers.py | 2 ++ 3 files changed, 42 insertions(+) diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index db9e3e4c711..787c6b66367 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -48060,6 +48060,26 @@ "supports_response_schema": true, "supports_tool_choice": true }, + "runinfra/deepseek-ai/DeepSeek-V4-Pro-0813": { + "cache_read_input_token_cost": 3e-08, + "input_cost_per_token": 6e-07, + "litellm_provider": "runinfra", + "max_input_tokens": 1048576, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 1.9e-06, + "source": "https://runinfra.ai/inference-api/deepseek-v4-pro", + "supported_endpoints": [ + "/v1/chat/completions" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_tool_choice": true + }, "runinfra/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16": { "input_cost_per_token": 5e-08, "litellm_provider": "runinfra", diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index db9e3e4c711..787c6b66367 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -48060,6 +48060,26 @@ "supports_response_schema": true, "supports_tool_choice": true }, + "runinfra/deepseek-ai/DeepSeek-V4-Pro-0813": { + "cache_read_input_token_cost": 3e-08, + "input_cost_per_token": 6e-07, + "litellm_provider": "runinfra", + "max_input_tokens": 1048576, + "max_output_tokens": 32768, + "max_tokens": 32768, + "mode": "chat", + "output_cost_per_token": 1.9e-06, + "source": "https://runinfra.ai/inference-api/deepseek-v4-pro", + "supported_endpoints": [ + "/v1/chat/completions" + ], + "supports_function_calling": true, + "supports_native_streaming": true, + "supports_prompt_caching": true, + "supports_reasoning": true, + "supports_response_schema": true, + "supports_tool_choice": true + }, "runinfra/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16": { "input_cost_per_token": 5e-08, "litellm_provider": "runinfra", diff --git a/tests/test_litellm/llms/openai_like/test_json_providers.py b/tests/test_litellm/llms/openai_like/test_json_providers.py index 7c9b2f1a98b..bf27307d8fc 100644 --- a/tests/test_litellm/llms/openai_like/test_json_providers.py +++ b/tests/test_litellm/llms/openai_like/test_json_providers.py @@ -419,6 +419,7 @@ class TestRuninfra: expected_models = { "runinfra/deepseek-ai/DeepSeek-V4-Flash-0731": (1.3e-07, 2.7e-07, 1e-08), + "runinfra/deepseek-ai/DeepSeek-V4-Pro-0813": (6e-07, 1.9e-06, 3e-08), "runinfra/nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16": ( 5e-08, 1.5e-07, @@ -444,6 +445,7 @@ class TestRuninfra: assert model_cost[model]["supports_prompt_caching"] is True assert model_cost["runinfra/deepseek-ai/DeepSeek-V4-Flash-0731"]["max_input_tokens"] == 1048576 + assert model_cost["runinfra/deepseek-ai/DeepSeek-V4-Pro-0813"]["max_input_tokens"] == 1048576 assert model_cost["runinfra/Qwen/Qwen3.8-27B"]["max_input_tokens"] == 262144 assert model_cost["runinfra/Inferact/Qwen3.8-2.4T-A95B-NVFP4"]["supports_response_schema"] is True assert model_cost["runinfra/Qwen/Qwen3.8-27B"]["supports_response_schema"] is True