diff --git a/providers/prism/logo.svg b/providers/prism/logo.svg
new file mode 100644
index 00000000000..0a31065db9e
--- /dev/null
+++ b/providers/prism/logo.svg
@@ -0,0 +1,4 @@
+<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 630 630" fill="currentColor">
+  <path d="M2 581 314 45l101 188-276 349Z"/>
+  <path d="m340 580 98-319 190 328Z"/>
+</svg>
diff --git a/providers/prism/models/prism/deepseek-v4-flash.toml b/providers/prism/models/prism/deepseek-v4-flash.toml
new file mode 100644
index 00000000000..2a547e723b8
--- /dev/null
+++ b/providers/prism/models/prism/deepseek-v4-flash.toml
@@ -0,0 +1,21 @@
+# Sources:
+# https://prisminference.com/pricing
+# https://docs.prisminference.com/models
+# https://docs.prisminference.com/api-reference/chat-completions
+# Reasoning: reasoning_effort = none|low|medium|high; none disables reasoning.
+base_model = "deepseek/deepseek-v4-flash"
+
+[interleaved]
+field = "reasoning_content"
+
+[[reasoning_options]]
+type = "effort"
+values = ["none", "low", "medium", "high"]
+
+[cost]
+input = 0.14
+output = 0.28
+cache_read = 0.07
+
+[limit]
+input = 1_000_000
diff --git a/providers/prism/models/prism/deepseek-v4.1-flash.toml b/providers/prism/models/prism/deepseek-v4.1-flash.toml
new file mode 100644
index 00000000000..9d93733c4b2
--- /dev/null
+++ b/providers/prism/models/prism/deepseek-v4.1-flash.toml
@@ -0,0 +1,21 @@
+# Sources:
+# https://prisminference.com/pricing
+# https://docs.prisminference.com/models
+# https://docs.prisminference.com/api-reference/chat-completions
+# Reasoning: reasoning_effort = none|low|medium|high; none disables reasoning.
+base_model = "deepseek/deepseek-v4.1-flash"
+
+[interleaved]
+field = "reasoning_content"
+
+[[reasoning_options]]
+type = "effort"
+values = ["none", "low", "medium", "high"]
+
+[cost]
+input = 0.30
+output = 1.20
+cache_read = 0.07
+
+[limit]
+input = 1_000_000
diff --git a/providers/prism/provider.toml b/providers/prism/provider.toml
new file mode 100644
index 00000000000..a4c4fc41899
--- /dev/null
+++ b/providers/prism/provider.toml
@@ -0,0 +1,13 @@
+# Raw HTTP routes are POST `/v1/chat/completions`, `/v1/responses`, and
+# `/v1/messages`. Chat uses `reasoning_effort = none|low|medium|high`, Responses
+# uses `reasoning.effort` with the same values, and Messages uses `thinking.type`
+# set to `enabled` or `disabled` with optional `budget_tokens` when enabled.
+# https://docs.prisminference.com/api-reference/chat-completions
+# https://docs.prisminference.com/api-reference/responses
+# https://docs.prisminference.com/guides/coding-agents
+# https://docs.prisminference.com/openapi.yaml
+name = "Prism"
+npm = "@ai-sdk/openai-compatible"
+env = ["PRISM_API_KEY"]
+api = "https://api.prisminference.com/v1"
+doc = "https://docs.prisminference.com/models"