Skip to content

AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.

fix(llm-registry): correct deepseek-v4.1-flash pricing drift (again) (#14856)

Nicholas Tindle02ccbdc4c4a2
diff --git a/autogpt_platform/backend/backend/data/llm_registry/catalog.py b/autogpt_platform/backend/backend/data/llm_registry/catalog.py
index b9a762e1ff..ad14325d5d 100644
--- a/autogpt_platform/backend/backend/data/llm_registry/catalog.py
+++ b/autogpt_platform/backend/backend/data/llm_registry/catalog.py
@@ -529,12 +529,12 @@ def _build_catalog() -> CatalogPayload:
context_window=1048576,
max_output_tokens=384000,
price_tier=1,
- # Live OpenRouter rate as of 2026-09-23: $0.06/$0.32 per 1M,
- # cache read $0.01/1M (further drift since this PR's prior fix).
+ # Live OpenRouter rate as of 2026-09-24: $0.14/$0.42 per 1M,
+ # cache read $0.01/1M (drifted again since the prior fix).
cost=CatalogModelCost(
run_credits=1,
- input_credits_per_1m=9.0,
- output_credits_per_1m=48.0,
+ input_credits_per_1m=21.0,
+ output_credits_per_1m=63.0,
cache_read_credits_per_1m=1.5,
),
),
diff --git a/autogpt_platform/backend/backend/data/llm_registry/catalog_test.py b/autogpt_platform/backend/backend/data/llm_registry/catalog_test.py
index 45570aa415..dc96a34902 100644
--- a/autogpt_platform/backend/backend/data/llm_registry/catalog_test.py
+++ b/autogpt_platform/backend/backend/data/llm_registry/catalog_test.py
@@ -394,14 +394,14 @@ def test_qwen3_8_flash_bills_at_authored_rates():
def test_deepseek_v4_1_flash_bills_at_authored_rates():
- """DeepSeek V4.1 Flash (OpenRouter live rate $0.06/$0.32 per 1M,
- $0.01/1M cached input as of 2026-09-23) — flat tier and per-1M
+ """DeepSeek V4.1 Flash (OpenRouter live rate $0.14/$0.42 per 1M,
+ $0.01/1M cached input as of 2026-09-24) — flat tier and per-1M
projections must match the authored catalog entry."""
flash = LLMModel("deepseek/deepseek-v4.1-flash")
assert MODEL_COST[flash] == 1
assert TOKEN_COST[flash].model_dump() == {
- "input": 9.0,
- "output": 48.0,
+ "input": 21.0,
+ "output": 63.0,
"cache_read": 1.5,
"cache_creation": 0.0,
}