feat(ollama-cloud): add deepseek-v4-flash:0731 model (#3985)

2026-08-03 16:44:27 +02:00
parent d5931d97c2
commit 0ccae5d09e
1 changed files with 12 additions and 0 deletions
@@ -0,0 +1,12 @@
+# Toggle: native POST /api/chat and /api/generate use `think = true|false`.
+# Effort: OpenAI POST /v1/chat/completions accepts top-level `reasoning_effort`
+# or `reasoning.effort` with high|max (DeepSeek V4 maps low/medium -> high).
+# Sources: https://docs.ollama.com/capabilities/thinking, https://docs.ollama.com/openai
+# API-verified context: https://ollama.com/api/show (deepseek4.context_length = 1048576)
+base_model = "deepseek/deepseek-v4-flash-0731"
+
+reasoning_options = [{ type = "toggle" }, { type = "effort", values = ["high", "max"] }]
+
+[limit]
+context = 1048576
+output = 1048576