From a47c0a8383af0b0a313df23fbb2f8940fb99c92d Mon Sep 17 00:00:00 2001 From: Jaber Jaber <103749727+jaberjaber23@users.noreply.github.com> Date: Mon, 19 Oct 2026 17:26:01 -0700 Subject: [PATCH] feat(runinfra): GLM-5.3 at its new standing rates RunInfra lowered GLM-5.3 on 2026-10-20 to $1.30 input, $4.20 output and $0.13 cached input per million tokens (1.3e-06, 4.2e-06 and 1.3e-07 per token). The model page and RunInfra's provider feed carry the same prices. Co-Authored-By: Claude Opus 5.5 --- litellm/model_prices_and_context_window_backup.json | 6 +++--- model_prices_and_context_window.json | 6 +++--- tests/unit/llms/openai_like/test_json_providers.py | 2 +- 3 files changed, 7 insertions(+), 7 deletions(-) diff --git a/litellm/model_prices_and_context_window_backup.json b/litellm/model_prices_and_context_window_backup.json index a6f41aaaae9..e4458b104ac 100644 --- a/litellm/model_prices_and_context_window_backup.json +++ b/litellm/model_prices_and_context_window_backup.json @@ -62487,14 +62487,14 @@ "supports_vision": true }, "runinfra/zai-org/GLM-5.3": { - "cache_read_input_token_cost": 1.4e-07, - "input_cost_per_token": 1.4e-06, + "cache_read_input_token_cost": 1.3e-07, + "input_cost_per_token": 1.3e-06, "litellm_provider": "runinfra", "max_input_tokens": 1048576, "max_output_tokens": 32768, "max_tokens": 32768, "mode": "chat", - "output_cost_per_token": 4.4e-06, + "output_cost_per_token": 4.2e-06, "source": "https://runinfra.ai/inference-api/glm-5-3", "supported_endpoints": [ "/v1/chat/completions" diff --git a/model_prices_and_context_window.json b/model_prices_and_context_window.json index a6f41aaaae9..e4458b104ac 100644 --- a/model_prices_and_context_window.json +++ b/model_prices_and_context_window.json @@ -62487,14 +62487,14 @@ "supports_vision": true }, "runinfra/zai-org/GLM-5.3": { - "cache_read_input_token_cost": 1.4e-07, - "input_cost_per_token": 1.4e-06, + "cache_read_input_token_cost": 1.3e-07, + "input_cost_per_token": 1.3e-06, "litellm_provider": "runinfra", "max_input_tokens": 1048576, "max_output_tokens": 32768, "max_tokens": 32768, "mode": "chat", - "output_cost_per_token": 4.4e-06, + "output_cost_per_token": 4.2e-06, "source": "https://runinfra.ai/inference-api/glm-5-3", "supported_endpoints": [ "/v1/chat/completions" diff --git a/tests/unit/llms/openai_like/test_json_providers.py b/tests/unit/llms/openai_like/test_json_providers.py index 2ae049d889b..6dee35d8ac0 100644 --- a/tests/unit/llms/openai_like/test_json_providers.py +++ b/tests/unit/llms/openai_like/test_json_providers.py @@ -402,7 +402,7 @@ class TestRuninfra: "runinfra/Qwen/Qwen3.8-27B": (1e-07, 4e-07, 1e-08), "runinfra/ornith-ai/Ornith-1.5-35B-A3B": (1e-07, 4e-07, 1e-08), "runinfra/zai-org/GLM-5.3-Flash": (1.1e-07, 4.5e-07, 3e-08), - "runinfra/zai-org/GLM-5.3": (1.4e-06, 4.4e-06, 1.4e-07), + "runinfra/zai-org/GLM-5.3": (1.3e-06, 4.2e-06, 1.3e-07), } # Every RunInfra entry is pinned here, so a retired model left in the # cost map, or a new one added without its prices, reddens this test.