Retry connection errors but never re-bill timeouts

num_retries=2 (retries transient connection errors), retry_policy with
TimeoutErrorRetries=0 (a timed-out generation is already billed by Tinfoil, so
re-hitting it re-bills). Combined with timeout=180, this hedges connection
errors without the silent re-billing that caused the Tinfoil spend gap.

config.yaml

@@ -49,13 +49,16 @@ model_list:
 # NOTE: DeepSeek V4.1 Flash is a reasoning model — generations routinely run
 # longer than 30s. A 30s timeout caused LiteLLM to abandon in-flight DeepSeek
 # requests (falling back to GPT-OSS), while Tinfoil still completed and billed
-# the abandoned generation. num_retries amplified this by re-hitting Tinfoil up
+# the abandoned generation. Result: Tinfoil billed ~2x the tokens LiteLLM logged.
-# to 3x per request. Result: Tinfoil billed ~2x the tokens LiteLLM logged.
+# Fix: a generous timeout (180s) so generations complete inside LiteLLM and get
-# Fix: a generous timeout so generations complete inside LiteLLM (and get
+# logged; and a retry_policy that retries transient connection errors (base
-# logged), and zero retries so a slow/failed request is never re-billed.
+# num_retries) but NEVER retries timeouts (TimeoutErrorRetries: 0) — a timed-out
+# generation is already billed by Tinfoil, so re-hitting it re-bills.
 router_settings:
-  num_retries: 0
+  num_retries: 2
   timeout: 180
+  retry_policy:
+    TimeoutErrorRetries: 0
   fallbacks:
     - deepseek-v4-1-flash:
         - gpt-oss-120b