diff --git a/specification/DigitalOcean-public.v2.yaml b/specification/DigitalOcean-public.v2.yaml
index ea5fca026..3c68c1bd1 100644
--- a/specification/DigitalOcean-public.v2.yaml
+++ b/specification/DigitalOcean-public.v2.yaml
@@ -3981,6 +3981,23 @@ paths:
     post:
       $ref: 'resources/inference/inference_async_invoke.yml'
 
+
+  /v1/audio/speech:
+    post:
+      $ref: 'resources/inference/inference_create_speech.yml'
+
+  /v1/videos:
+    post:
+      $ref: 'resources/inference/inference_create_video.yml'
+
+  /v1/videos/{video_id}:
+    get:
+      $ref: 'resources/inference/inference_get_video.yml'
+
+  /v1/videos/{video_id}/content:
+    get:
+      $ref: 'resources/inference/inference_get_video_content.yml'
+
   /v1/batches/files:
     post:
       $ref: 'resources/inference/inference_create_batch_file.yml'
diff --git a/specification/resources/inference/examples/curl/inference_create_speech.yml b/specification/resources/inference/examples/curl/inference_create_speech.yml
new file mode 100644
index 000000000..3ab78b694
--- /dev/null
+++ b/specification/resources/inference/examples/curl/inference_create_speech.yml
@@ -0,0 +1,13 @@
+lang: cURL
+source: |-
+  curl -sS https://inference.do-ai.run/v1/audio/speech \
+    -H "Authorization: Bearer $DIGITALOCEAN_TOKEN" \
+    -H "Content-Type: application/json" \
+    -d '{
+      "model": "qwen3-tts-voicedesign",
+      "input": "Hello, world!",
+      "voice": "alloy",
+      "response_format": "mp3",
+      "instructions": "Speak naturally."
+    }' \
+    --output output.mp3
diff --git a/specification/resources/inference/examples/curl/inference_create_video.yml b/specification/resources/inference/examples/curl/inference_create_video.yml
new file mode 100644
index 000000000..d3b1913ba
--- /dev/null
+++ b/specification/resources/inference/examples/curl/inference_create_video.yml
@@ -0,0 +1,11 @@
+lang: cURL
+source: |-
+  curl -X POST https://inference.do-ai.run/v1/videos \
+    -H "Authorization: Bearer $DIGITALOCEAN_TOKEN" \
+    -H "Content-Type: application/json" \
+    -d '{
+      "model": "wan2-2-t2v-a14b",
+      "prompt": "Two anthropomorphic cats in boxing gear fight intensely on a spotlighted stage",
+      "size": "1280x720",
+      "fps": 16
+    }'
diff --git a/specification/resources/inference/examples/curl/inference_get_video.yml b/specification/resources/inference/examples/curl/inference_get_video.yml
new file mode 100644
index 000000000..e77bcda2e
--- /dev/null
+++ b/specification/resources/inference/examples/curl/inference_get_video.yml
@@ -0,0 +1,4 @@
+lang: cURL
+source: |-
+  curl https://inference.do-ai.run/v1/videos/$VIDEO_ID \
+    -H "Authorization: Bearer $DIGITALOCEAN_TOKEN"
diff --git a/specification/resources/inference/examples/curl/inference_get_video_content.yml b/specification/resources/inference/examples/curl/inference_get_video_content.yml
new file mode 100644
index 000000000..5bfd35e63
--- /dev/null
+++ b/specification/resources/inference/examples/curl/inference_get_video_content.yml
@@ -0,0 +1,5 @@
+lang: cURL
+source: |-
+  curl https://inference.do-ai.run/v1/videos/$VIDEO_ID/content \
+    -H "Authorization: Bearer $DIGITALOCEAN_TOKEN" \
+    --output output.mp4
diff --git a/specification/resources/inference/examples/dots/inference_create_speech.yml b/specification/resources/inference/examples/dots/inference_create_speech.yml
new file mode 100644
index 000000000..6364f2dc9
--- /dev/null
+++ b/specification/resources/inference/examples/dots/inference_create_speech.yml
@@ -0,0 +1,19 @@
+lang: JavaScript
+source: |-
+  import { writeFileSync } from "fs";
+  import { InferenceClient } from "@digitalocean/dots";
+
+  const client = new InferenceClient({
+      apiKey: process.env.DIGITALOCEAN_TOKEN,
+  });
+
+  const response = await client.audio.speech.create({
+      model: "qwen3-tts-voicedesign",
+      input: "Hello, world!",
+      voice: "alloy",
+      response_format: "mp3",
+      instructions: "Speak naturally.",
+  });
+
+  const audioBytes = Buffer.from(response.data[0].b64_json, "base64");
+  writeFileSync("output.mp3", audioBytes);
diff --git a/specification/resources/inference/examples/dots/inference_create_video.yml b/specification/resources/inference/examples/dots/inference_create_video.yml
new file mode 100644
index 000000000..4212f19c4
--- /dev/null
+++ b/specification/resources/inference/examples/dots/inference_create_video.yml
@@ -0,0 +1,16 @@
+lang: JavaScript
+source: |-
+  import { InferenceClient } from "@digitalocean/dots";
+
+  const client = new InferenceClient({
+      apiKey: process.env.DIGITALOCEAN_TOKEN,
+  });
+
+  const job = await client.video.generations.create({
+      model: "wan2-2-t2v-a14b",
+      prompt: "Two anthropomorphic cats in boxing gear fight intensely on a spotlighted stage",
+      size: "1280x720",
+      fps: 16,
+  });
+
+  console.log(job.id, job.status);
diff --git a/specification/resources/inference/examples/dots/inference_get_video.yml b/specification/resources/inference/examples/dots/inference_get_video.yml
new file mode 100644
index 000000000..322b07ba9
--- /dev/null
+++ b/specification/resources/inference/examples/dots/inference_get_video.yml
@@ -0,0 +1,10 @@
+lang: JavaScript
+source: |-
+  import { InferenceClient } from "@digitalocean/dots";
+
+  const client = new InferenceClient({
+      apiKey: process.env.DIGITALOCEAN_TOKEN,
+  });
+
+  const job = await client.video.generations.retrieve(process.env.VIDEO_ID);
+  console.log(job.id, job.status);
diff --git a/specification/resources/inference/examples/dots/inference_get_video_content.yml b/specification/resources/inference/examples/dots/inference_get_video_content.yml
new file mode 100644
index 000000000..a2e4e31db
--- /dev/null
+++ b/specification/resources/inference/examples/dots/inference_get_video_content.yml
@@ -0,0 +1,11 @@
+lang: JavaScript
+source: |-
+  import { writeFileSync } from "fs";
+  import { InferenceClient } from "@digitalocean/dots";
+
+  const client = new InferenceClient({
+      apiKey: process.env.DIGITALOCEAN_TOKEN,
+  });
+
+  const content = await client.video.generations.content(process.env.VIDEO_ID);
+  writeFileSync("output.mp4", Buffer.from(await content.arrayBuffer()));
diff --git a/specification/resources/inference/examples/go/agentInference_create_chat_completion.yml b/specification/resources/inference/examples/go/agentInference_create_chat_completion.yml
index aece2c5b9..d00fba084 100644
--- a/specification/resources/inference/examples/go/agentInference_create_chat_completion.yml
+++ b/specification/resources/inference/examples/go/agentInference_create_chat_completion.yml
@@ -32,6 +32,6 @@ source: |-
       }
 
       if msg := completion.Choices[0].Message.Content; msg != nil {
-          fmt.Println(*msg)
+          fmt.Println(msg.Text())
       }
   }
diff --git a/specification/resources/inference/examples/go/inference_create_chat_completion.yml b/specification/resources/inference/examples/go/inference_create_chat_completion.yml
index 95bb3dc9c..2d9fc15a3 100644
--- a/specification/resources/inference/examples/go/inference_create_chat_completion.yml
+++ b/specification/resources/inference/examples/go/inference_create_chat_completion.yml
@@ -25,6 +25,6 @@ source: |-
       }
 
       if msg := completion.Choices[0].Message.Content; msg != nil {
-          fmt.Println(*msg)
+          fmt.Println(msg.Text())
       }
   }
diff --git a/specification/resources/inference/examples/go/inference_create_chat_completion_kb.yml b/specification/resources/inference/examples/go/inference_create_chat_completion_kb.yml
new file mode 100644
index 000000000..bf7f3fc30
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_create_chat_completion_kb.yml
@@ -0,0 +1,37 @@
+lang: Go
+label: Knowledge base
+source: |-
+  package main
+
+  import (
+      "context"
+      "encoding/json"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      completion, _, err := client.Chat.Completions.New(ctx, &godo.ChatCompletionNewParams{
+          Model: "openai-gpt-oss-20b",
+          Messages: []godo.ChatCompletionMessage{
+              godo.UserMessage("What are some features of DigitalOcean Inference?"),
+          },
+          Tools: []godo.ChatCompletionTool{
+              godo.ChatKnowledgeBaseRetrievalTool(os.Getenv("KNOWLEDGE_BASE_ID")),
+          },
+          ToolChoice: json.RawMessage(`"auto"`),
+          MaxTokens:  godo.PtrTo(1024),
+      })
+      if err != nil {
+          panic(err)
+      }
+
+      if msg := completion.Choices[0].Message.Content; msg != nil {
+          fmt.Println(msg.Text())
+      }
+  }
diff --git a/specification/resources/inference/examples/go/inference_create_chat_completion_mcp.yml b/specification/resources/inference/examples/go/inference_create_chat_completion_mcp.yml
new file mode 100644
index 000000000..88d8d4c05
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_create_chat_completion_mcp.yml
@@ -0,0 +1,42 @@
+lang: Go
+label: MCP tool
+source: |-
+  package main
+
+  import (
+      "context"
+      "encoding/json"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      completion, _, err := client.Chat.Completions.New(ctx, &godo.ChatCompletionNewParams{
+          Model: "openai-gpt-oss-20b",
+          Messages: []godo.ChatCompletionMessage{
+              godo.UserMessage("Fetch my DigitalOcean account information and summarize it in 2 bullets."),
+          },
+          Tools: []godo.ChatCompletionTool{
+              godo.ChatMCPTool(godo.MCPToolParams{
+                  ServerLabel:   "digitalocean",
+                  ServerURL:     "https://accounts.mcp.digitalocean.com/mcp",
+                  Authorization: "Bearer " + os.Getenv("DIGITALOCEAN_API_TOKEN"),
+                  AllowedTools:  []string{"account-get-information"},
+              }),
+          },
+          ToolChoice: json.RawMessage(`"required"`),
+          MaxTokens:  godo.PtrTo(512),
+      })
+      if err != nil {
+          panic(err)
+      }
+
+      if msg := completion.Choices[0].Message.Content; msg != nil {
+          fmt.Println(msg.Text())
+      }
+  }
diff --git a/specification/resources/inference/examples/go/inference_create_chat_completion_vlm.yml b/specification/resources/inference/examples/go/inference_create_chat_completion_vlm.yml
new file mode 100644
index 000000000..58439273b
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_create_chat_completion_vlm.yml
@@ -0,0 +1,41 @@
+lang: Go
+label: Vision (VLM)
+source: |-
+  package main
+
+  import (
+      "context"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      // IMAGE_URL may be an HTTPS URL or a data:image/...;base64,... URI.
+      imageURL := os.Getenv("IMAGE_URL")
+      if imageURL == "" {
+          imageURL = "https://upload.wikimedia.org/wikipedia/commons/thumb/3/3b/Online_Bicycles.jpg/640px-Online_Bicycles.jpg"
+      }
+
+      completion, _, err := client.Chat.Completions.New(ctx, &godo.ChatCompletionNewParams{
+          Model: "nemotron-nano-12b-v2-vl",
+          Messages: []godo.ChatCompletionMessage{
+              godo.UserMessageParts(
+                  godo.TextPart("Describe this image."),
+                  godo.ImageURLPart(imageURL),
+              ),
+          },
+          MaxTokens: godo.PtrTo(512),
+      })
+      if err != nil {
+          panic(err)
+      }
+
+      if msg := completion.Choices[0].Message.Content; msg != nil {
+          fmt.Println(msg.Text())
+      }
+  }
diff --git a/specification/resources/inference/examples/go/inference_create_response_model_synthesis.yml b/specification/resources/inference/examples/go/inference_create_response_model_synthesis.yml
new file mode 100644
index 000000000..49b92b9db
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_create_response_model_synthesis.yml
@@ -0,0 +1,33 @@
+lang: Go
+label: Model synthesis
+source: |-
+  package main
+
+  import (
+      "context"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      resp, _, err := client.Responses.New(ctx, &godo.ResponseNewParams{
+          Model: "openai-gpt-oss-20b",
+          Input: "Compare approaches for reducing inference latency.",
+          Tools: []godo.ResponseTool{
+              godo.ModelSynthesisTool(godo.ModelSynthesisToolParams{
+                  Preset: "balanced",
+              }),
+          },
+          MaxOutputTokens: godo.PtrTo(1024),
+      })
+      if err != nil {
+          panic(err)
+      }
+
+      fmt.Println(resp.OutputText())
+  }
diff --git a/specification/resources/inference/examples/go/inference_create_response_tool_search.yml b/specification/resources/inference/examples/go/inference_create_response_tool_search.yml
new file mode 100644
index 000000000..40ac3c803
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_create_response_tool_search.yml
@@ -0,0 +1,38 @@
+lang: Go
+label: Tool search
+source: |-
+  package main
+
+  import (
+      "context"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      resp, _, err := client.Responses.New(ctx, &godo.ResponseNewParams{
+          Model: "openai-gpt-5.5",
+          Input: "Compare the current weather in zip code 94107 and 10001.",
+          Tools: []godo.ResponseTool{
+              godo.FunctionTool("get_weather", "Get weather for a zip code", map[string]interface{}{
+                  "type": "object",
+                  "properties": map[string]interface{}{
+                      "zip": map[string]interface{}{"type": "string"},
+                  },
+                  "required": []string{"zip"},
+              }),
+              godo.ToolSearchTool(),
+          },
+          MaxOutputTokens: godo.PtrTo(1024),
+      })
+      if err != nil {
+          panic(err)
+      }
+
+      fmt.Println(resp.OutputText())
+  }
diff --git a/specification/resources/inference/examples/go/inference_create_response_web_fetch.yml b/specification/resources/inference/examples/go/inference_create_response_web_fetch.yml
new file mode 100644
index 000000000..7d7130537
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_create_response_web_fetch.yml
@@ -0,0 +1,31 @@
+lang: Go
+label: Web fetch
+source: |-
+  package main
+
+  import (
+      "context"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      resp, _, err := client.Responses.New(ctx, &godo.ResponseNewParams{
+          Model: "openai-gpt-oss-20b",
+          Input: "Summarize https://www.digitalocean.com/pricing",
+          Tools: []godo.ResponseTool{
+              godo.WebFetchTool(godo.PtrTo(3)),
+          },
+          MaxOutputTokens: godo.PtrTo(1024),
+      })
+      if err != nil {
+          panic(err)
+      }
+
+      fmt.Println(resp.OutputText())
+  }
diff --git a/specification/resources/inference/examples/go/inference_create_response_web_search.yml b/specification/resources/inference/examples/go/inference_create_response_web_search.yml
new file mode 100644
index 000000000..6a3486f8f
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_create_response_web_search.yml
@@ -0,0 +1,31 @@
+lang: Go
+label: Web search
+source: |-
+  package main
+
+  import (
+      "context"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      resp, _, err := client.Responses.New(ctx, &godo.ResponseNewParams{
+          Model: "openai-gpt-oss-20b",
+          Input: "What are the latest pricing changes for DigitalOcean Droplets?",
+          Tools: []godo.ResponseTool{
+              godo.WebSearchTool(godo.PtrTo(3), godo.PtrTo(5)),
+          },
+          MaxOutputTokens: godo.PtrTo(1024),
+      })
+      if err != nil {
+          panic(err)
+      }
+
+      fmt.Println(resp.OutputText())
+  }
diff --git a/specification/resources/inference/examples/go/inference_create_speech.yml b/specification/resources/inference/examples/go/inference_create_speech.yml
new file mode 100644
index 000000000..03b2e1b0c
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_create_speech.yml
@@ -0,0 +1,36 @@
+lang: Go
+source: |-
+  package main
+
+  import (
+      "context"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      speech, _, err := client.Audio.Speech.New(ctx, &godo.SpeechNewParams{
+          Model:          "qwen3-tts-voicedesign",
+          Input:          "Hello, world!",
+          Voice:          "alloy",
+          ResponseFormat: "mp3",
+          Instructions:   "Speak naturally.",
+      })
+      if err != nil {
+          panic(err)
+      }
+
+      audioBytes, err := speech.Bytes()
+      if err != nil {
+          panic(err)
+      }
+      if err := os.WriteFile("output.mp3", audioBytes, 0644); err != nil {
+          panic(err)
+      }
+      fmt.Println("Wrote output.mp3")
+  }
diff --git a/specification/resources/inference/examples/go/inference_create_video.yml b/specification/resources/inference/examples/go/inference_create_video.yml
new file mode 100644
index 000000000..1457a0bf2
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_create_video.yml
@@ -0,0 +1,28 @@
+lang: Go
+source: |-
+  package main
+
+  import (
+      "context"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      job, _, err := client.Video.Generations.Create(ctx, &godo.VideoGenerateParams{
+          Model:  "wan2-2-t2v-a14b",
+          Prompt: "Two anthropomorphic cats in boxing gear fight intensely on a spotlighted stage",
+          Size:   "1280x720",
+          FPS:    godo.PtrTo(16),
+      })
+      if err != nil {
+          panic(err)
+      }
+
+      fmt.Println(job.ID, job.Status)
+  }
diff --git a/specification/resources/inference/examples/go/inference_get_video.yml b/specification/resources/inference/examples/go/inference_get_video.yml
new file mode 100644
index 000000000..6a2458c03
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_get_video.yml
@@ -0,0 +1,23 @@
+lang: Go
+source: |-
+  package main
+
+  import (
+      "context"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      job, _, err := client.Video.Generations.Get(ctx, os.Getenv("VIDEO_ID"))
+      if err != nil {
+          panic(err)
+      }
+
+      fmt.Println(job.ID, job.Status)
+  }
diff --git a/specification/resources/inference/examples/go/inference_get_video_content.yml b/specification/resources/inference/examples/go/inference_get_video_content.yml
new file mode 100644
index 000000000..601bfdac7
--- /dev/null
+++ b/specification/resources/inference/examples/go/inference_get_video_content.yml
@@ -0,0 +1,25 @@
+lang: Go
+source: |-
+  package main
+
+  import (
+      "context"
+      "fmt"
+      "os"
+
+      "github.com/digitalocean/godo"
+  )
+
+  func main() {
+      client := godo.NewFromToken(os.Getenv("DIGITALOCEAN_TOKEN"))
+      ctx := context.Background()
+
+      raw, _, err := client.Video.Generations.Content(ctx, os.Getenv("VIDEO_ID"))
+      if err != nil {
+          panic(err)
+      }
+      if err := os.WriteFile("output.mp4", raw, 0644); err != nil {
+          panic(err)
+      }
+      fmt.Println("Wrote output.mp4")
+  }
diff --git a/specification/resources/inference/examples/python/inference_create_speech.yml b/specification/resources/inference/examples/python/inference_create_speech.yml
new file mode 100644
index 000000000..46e28a161
--- /dev/null
+++ b/specification/resources/inference/examples/python/inference_create_speech.yml
@@ -0,0 +1,19 @@
+lang: Python
+source: |-
+  import base64
+  import os
+  from pydo.inference import Client
+
+  client = Client(token=os.environ.get("DIGITALOCEAN_TOKEN"))
+
+  response = client.audio.speech.create(
+      model="qwen3-tts-voicedesign",
+      input="Hello, world!",
+      voice="alloy",
+      response_format="mp3",
+      instructions="Speak naturally.",
+  )
+
+  audio_bytes = base64.b64decode(response.data[0].b64_json)
+  with open("output.mp3", "wb") as f:
+      f.write(audio_bytes)
diff --git a/specification/resources/inference/examples/python/inference_create_video.yml b/specification/resources/inference/examples/python/inference_create_video.yml
new file mode 100644
index 000000000..2402b6fce
--- /dev/null
+++ b/specification/resources/inference/examples/python/inference_create_video.yml
@@ -0,0 +1,15 @@
+lang: Python
+source: |-
+  import os
+  from pydo.inference import Client
+
+  client = Client(token=os.environ.get("DIGITALOCEAN_TOKEN"))
+
+  job = client.video.generations.create(
+      model="wan2-2-t2v-a14b",
+      prompt="Two anthropomorphic cats in boxing gear fight intensely on a spotlighted stage",
+      size="1280x720",
+      fps=16,
+  )
+
+  print(job.id, job.status)
diff --git a/specification/resources/inference/examples/python/inference_get_video.yml b/specification/resources/inference/examples/python/inference_get_video.yml
new file mode 100644
index 000000000..614d67400
--- /dev/null
+++ b/specification/resources/inference/examples/python/inference_get_video.yml
@@ -0,0 +1,9 @@
+lang: Python
+source: |-
+  import os
+  from pydo.inference import Client
+
+  client = Client(token=os.environ.get("DIGITALOCEAN_TOKEN"))
+
+  job = client.video.generations.retrieve(os.environ["VIDEO_ID"])
+  print(job.id, job.status)
diff --git a/specification/resources/inference/examples/python/inference_get_video_content.yml b/specification/resources/inference/examples/python/inference_get_video_content.yml
new file mode 100644
index 000000000..fef181f19
--- /dev/null
+++ b/specification/resources/inference/examples/python/inference_get_video_content.yml
@@ -0,0 +1,11 @@
+lang: Python
+source: |-
+  import os
+  import requests
+
+  headers = {"Authorization": f"Bearer {os.environ['DIGITALOCEAN_TOKEN']}"}
+  url = f"https://inference.do-ai.run/v1/videos/{os.environ['VIDEO_ID']}/content"
+  resp = requests.get(url, headers=headers)
+  resp.raise_for_status()
+  with open("output.mp4", "wb") as f:
+      f.write(resp.content)
diff --git a/specification/resources/inference/inference_create_chat_completion.yml b/specification/resources/inference/inference_create_chat_completion.yml
index 5fefca8c5..f8dc1de76 100644
--- a/specification/resources/inference/inference_create_chat_completion.yml
+++ b/specification/resources/inference/inference_create_chat_completion.yml
@@ -42,6 +42,9 @@ x-codeSamples:
   - $ref: 'examples/curl/inference_create_chat_completion.yml'
   - $ref: 'examples/python/inference_create_chat_completion.yml'
   - $ref: 'examples/go/inference_create_chat_completion.yml'
+  - $ref: 'examples/go/inference_create_chat_completion_mcp.yml'
+  - $ref: 'examples/go/inference_create_chat_completion_vlm.yml'
+  - $ref: 'examples/go/inference_create_chat_completion_kb.yml'
   - $ref: 'examples/dots/inference_create_chat_completion.yml'
 security:
   - inference_bearer_auth: []
diff --git a/specification/resources/inference/inference_create_response.yml b/specification/resources/inference/inference_create_response.yml
index 3250a5afd..ca3cc181a 100644
--- a/specification/resources/inference/inference_create_response.yml
+++ b/specification/resources/inference/inference_create_response.yml
@@ -43,6 +43,10 @@ x-codeSamples:
   - $ref: 'examples/curl/inference_create_response.yml'
   - $ref: 'examples/python/inference_create_response.yml'
   - $ref: 'examples/go/inference_create_response.yml'
+  - $ref: 'examples/go/inference_create_response_web_search.yml'
+  - $ref: 'examples/go/inference_create_response_web_fetch.yml'
+  - $ref: 'examples/go/inference_create_response_model_synthesis.yml'
+  - $ref: 'examples/go/inference_create_response_tool_search.yml'
   - $ref: 'examples/dots/inference_create_response.yml'
 security:
   - inference_bearer_auth: []
diff --git a/specification/resources/inference/inference_create_speech.yml b/specification/resources/inference/inference_create_speech.yml
new file mode 100644
index 000000000..ff5de6366
--- /dev/null
+++ b/specification/resources/inference/inference_create_speech.yml
@@ -0,0 +1,59 @@
+operationId: inference_create_speech
+summary: Generate speech from text
+description: >
+  Convert text input to speech audio. The response may be raw audio bytes
+  (`audio/mpeg`, `audio/wav`, and similar) or a JSON envelope with
+  base64-encoded audio in `data[].b64_json`.
+tags:
+  - Serverless Inference
+servers:
+  - url: "https://inference.do-ai.run"
+    description: production
+x-inference-base-url: "https://inference.do-ai.run"
+requestBody:
+  required: true
+  content:
+    application/json:
+      schema:
+        $ref: "models/create_speech_request.yml"
+responses:
+  "200":
+    description: Successful speech generation.
+    headers:
+      ratelimit-limit:
+        $ref: '../../shared/headers.yml#/ratelimit-limit'
+      ratelimit-remaining:
+        $ref: '../../shared/headers.yml#/ratelimit-remaining'
+      ratelimit-reset:
+        $ref: '../../shared/headers.yml#/ratelimit-reset'
+    content:
+      application/json:
+        schema:
+          $ref: "models/speech_response.yml"
+      audio/mpeg:
+        schema:
+          type: string
+          format: binary
+      audio/wav:
+        schema:
+          type: string
+          format: binary
+      audio/mp3:
+        schema:
+          type: string
+          format: binary
+  "401":
+    $ref: '../../shared/responses/unauthorized.yml'
+  "429":
+    $ref: '../../shared/responses/too_many_requests.yml'
+  "500":
+    $ref: '../../shared/responses/server_error.yml'
+  default:
+    $ref: '../../shared/responses/unexpected_error.yml'
+x-codeSamples:
+  - $ref: 'examples/curl/inference_create_speech.yml'
+  - $ref: 'examples/python/inference_create_speech.yml'
+  - $ref: 'examples/go/inference_create_speech.yml'
+  - $ref: 'examples/dots/inference_create_speech.yml'
+security:
+  - inference_bearer_auth: []
diff --git a/specification/resources/inference/inference_create_video.yml b/specification/resources/inference/inference_create_video.yml
new file mode 100644
index 000000000..1fa2036a3
--- /dev/null
+++ b/specification/resources/inference/inference_create_video.yml
@@ -0,0 +1,48 @@
+operationId: inference_create_video
+summary: Start a video generation job
+description: >
+  Start an asynchronous video generation job from a text prompt. Poll
+  `GET /v1/videos/{video_id}` until the job reaches a terminal status, then
+  download the MP4 with `GET /v1/videos/{video_id}/content` or from
+  `output.url` when present.
+tags:
+  - Serverless Inference
+servers:
+  - url: "https://inference.do-ai.run"
+    description: production
+x-inference-base-url: "https://inference.do-ai.run"
+requestBody:
+  required: true
+  content:
+    application/json:
+      schema:
+        $ref: "models/create_video_request.yml"
+responses:
+  "200":
+    description: Video generation job accepted.
+    headers:
+      ratelimit-limit:
+        $ref: '../../shared/headers.yml#/ratelimit-limit'
+      ratelimit-remaining:
+        $ref: '../../shared/headers.yml#/ratelimit-remaining'
+      ratelimit-reset:
+        $ref: '../../shared/headers.yml#/ratelimit-reset'
+    content:
+      application/json:
+        schema:
+          $ref: "models/video_generation.yml"
+  "401":
+    $ref: '../../shared/responses/unauthorized.yml'
+  "429":
+    $ref: '../../shared/responses/too_many_requests.yml'
+  "500":
+    $ref: '../../shared/responses/server_error.yml'
+  default:
+    $ref: '../../shared/responses/unexpected_error.yml'
+x-codeSamples:
+  - $ref: 'examples/curl/inference_create_video.yml'
+  - $ref: 'examples/python/inference_create_video.yml'
+  - $ref: 'examples/go/inference_create_video.yml'
+  - $ref: 'examples/dots/inference_create_video.yml'
+security:
+  - inference_bearer_auth: []
diff --git a/specification/resources/inference/inference_get_video.yml b/specification/resources/inference/inference_get_video.yml
new file mode 100644
index 000000000..5efeca715
--- /dev/null
+++ b/specification/resources/inference/inference_get_video.yml
@@ -0,0 +1,50 @@
+operationId: inference_get_video
+summary: Retrieve a video generation job
+description: >
+  Returns the current state of a video generation job. Poll until `status`
+  reaches a terminal value such as `completed` or `failed`.
+tags:
+  - Serverless Inference
+servers:
+  - url: "https://inference.do-ai.run"
+    description: production
+x-inference-base-url: "https://inference.do-ai.run"
+parameters:
+  - in: path
+    name: video_id
+    description: The video generation job identifier.
+    required: true
+    schema:
+      type: string
+    example: "video_bGl0ZWxsbTpjdXN0b21fbGxtX3Byb3ZpZGVyOm9wZW5haTttb2RlbF9pZDo7dmlkZW9faWQ6ZGExYjFmYzMtYTM3NC00MmM0LTgxMWMtOWE5NmE3ZWY0OTIz"
+responses:
+  "200":
+    description: The video generation job.
+    headers:
+      ratelimit-limit:
+        $ref: '../../shared/headers.yml#/ratelimit-limit'
+      ratelimit-remaining:
+        $ref: '../../shared/headers.yml#/ratelimit-remaining'
+      ratelimit-reset:
+        $ref: '../../shared/headers.yml#/ratelimit-reset'
+    content:
+      application/json:
+        schema:
+          $ref: "models/video_generation.yml"
+  "401":
+    $ref: '../../shared/responses/unauthorized.yml'
+  "404":
+    $ref: '../../shared/responses/not_found.yml'
+  "429":
+    $ref: '../../shared/responses/too_many_requests.yml'
+  "500":
+    $ref: '../../shared/responses/server_error.yml'
+  default:
+    $ref: '../../shared/responses/unexpected_error.yml'
+x-codeSamples:
+  - $ref: 'examples/curl/inference_get_video.yml'
+  - $ref: 'examples/python/inference_get_video.yml'
+  - $ref: 'examples/go/inference_get_video.yml'
+  - $ref: 'examples/dots/inference_get_video.yml'
+security:
+  - inference_bearer_auth: []
diff --git a/specification/resources/inference/inference_get_video_content.yml b/specification/resources/inference/inference_get_video_content.yml
new file mode 100644
index 000000000..c4f6ed300
--- /dev/null
+++ b/specification/resources/inference/inference_get_video_content.yml
@@ -0,0 +1,50 @@
+operationId: inference_get_video_content
+summary: Download generated video content
+description: >
+  Downloads the generated MP4 for a completed video generation job.
+tags:
+  - Serverless Inference
+servers:
+  - url: "https://inference.do-ai.run"
+    description: production
+x-inference-base-url: "https://inference.do-ai.run"
+parameters:
+  - in: path
+    name: video_id
+    description: The video generation job identifier.
+    required: true
+    schema:
+      type: string
+    example: "video_bGl0ZWxsbTpjdXN0b21fbGxtX3Byb3ZpZGVyOm9wZW5haTttb2RlbF9pZDo7dmlkZW9faWQ6ZGExYjFmYzMtYTM3NC00MmM0LTgxMWMtOWE5NmE3ZWY0OTIz"
+responses:
+  "200":
+    description: Binary MP4 content.
+    headers:
+      ratelimit-limit:
+        $ref: '../../shared/headers.yml#/ratelimit-limit'
+      ratelimit-remaining:
+        $ref: '../../shared/headers.yml#/ratelimit-remaining'
+      ratelimit-reset:
+        $ref: '../../shared/headers.yml#/ratelimit-reset'
+    content:
+      video/mp4:
+        schema:
+          type: string
+          format: binary
+  "401":
+    $ref: '../../shared/responses/unauthorized.yml'
+  "404":
+    $ref: '../../shared/responses/not_found.yml'
+  "429":
+    $ref: '../../shared/responses/too_many_requests.yml'
+  "500":
+    $ref: '../../shared/responses/server_error.yml'
+  default:
+    $ref: '../../shared/responses/unexpected_error.yml'
+x-codeSamples:
+  - $ref: 'examples/curl/inference_get_video_content.yml'
+  - $ref: 'examples/python/inference_get_video_content.yml'
+  - $ref: 'examples/go/inference_get_video_content.yml'
+  - $ref: 'examples/dots/inference_get_video_content.yml'
+security:
+  - inference_bearer_auth: []
diff --git a/specification/resources/inference/models/create_speech_request.yml b/specification/resources/inference/models/create_speech_request.yml
new file mode 100644
index 000000000..4e0f19e4f
--- /dev/null
+++ b/specification/resources/inference/models/create_speech_request.yml
@@ -0,0 +1,29 @@
+type: object
+description: Request body for text-to-speech generation.
+required:
+  - model
+  - input
+properties:
+  model:
+    type: string
+    description: The text-to-speech model to use.
+    example: "qwen3-tts-voicedesign"
+  input:
+    type: string
+    minLength: 1
+    description: The text to convert to speech.
+    example: "Hello, world!"
+  voice:
+    type: string
+    description: The voice to use for synthesis.
+    example: "alloy"
+  response_format:
+    type: string
+    description: >
+      The audio format to return. Supported values include `mp3` and `wav`.
+      The API may return raw audio bytes or a JSON envelope with base64-encoded audio.
+    example: "mp3"
+  instructions:
+    type: string
+    description: Optional style or delivery instructions for the synthesizer.
+    example: "Speak naturally."
diff --git a/specification/resources/inference/models/create_video_request.yml b/specification/resources/inference/models/create_video_request.yml
new file mode 100644
index 000000000..1535badb5
--- /dev/null
+++ b/specification/resources/inference/models/create_video_request.yml
@@ -0,0 +1,24 @@
+type: object
+description: Request body to start an asynchronous video generation job.
+required:
+  - model
+  - prompt
+properties:
+  model:
+    type: string
+    description: The video generation model to use.
+    example: "wan2-2-t2v-a14b"
+  prompt:
+    type: string
+    minLength: 1
+    description: A text description of the desired video.
+    example: "Two anthropomorphic cats in boxing gear fight intensely on a spotlighted stage"
+  size:
+    type: string
+    description: The output resolution, for example `1280x720`.
+    example: "1280x720"
+  fps:
+    type: integer
+    minimum: 1
+    description: Frames per second for the generated video.
+    example: 16
diff --git a/specification/resources/inference/models/speech_audio_item.yml b/specification/resources/inference/models/speech_audio_item.yml
new file mode 100644
index 000000000..860e269e1
--- /dev/null
+++ b/specification/resources/inference/models/speech_audio_item.yml
@@ -0,0 +1,9 @@
+type: object
+description: A single audio payload in a JSON speech response envelope.
+required:
+  - b64_json
+properties:
+  b64_json:
+    type: string
+    description: Base64-encoded audio bytes.
+    example: "//uQxAAAA..."
diff --git a/specification/resources/inference/models/speech_response.yml b/specification/resources/inference/models/speech_response.yml
new file mode 100644
index 000000000..28d6d9615
--- /dev/null
+++ b/specification/resources/inference/models/speech_response.yml
@@ -0,0 +1,17 @@
+type: object
+description: >
+  JSON envelope for text-to-speech when the endpoint returns `application/json`.
+  The live endpoint may instead return raw audio bytes with an `audio/*` Content-Type.
+required:
+  - data
+properties:
+  created:
+    type: integer
+    description: Unix timestamp (in seconds) of when the audio was created.
+    example: 1710700000
+  data:
+    type: array
+    minItems: 1
+    description: Generated audio items.
+    items:
+      $ref: "speech_audio_item.yml"
diff --git a/specification/resources/inference/models/video_generation.yml b/specification/resources/inference/models/video_generation.yml
new file mode 100644
index 000000000..0f6e36eeb
--- /dev/null
+++ b/specification/resources/inference/models/video_generation.yml
@@ -0,0 +1,40 @@
+type: object
+description: Status of an asynchronous video generation job.
+required:
+  - id
+  - status
+properties:
+  id:
+    type: string
+    description: The video generation job identifier.
+    example: "video_bGl0ZWxsbTpjdXN0b21fbGxtX3Byb3ZpZGVyOm9wZW5haTttb2RlbF9pZDo7dmlkZW9faWQ6ZGExYjFmYzMtYTM3NC00MmM0LTgxMWMtOWE5NmE3ZWY0OTIz"
+  object:
+    type: string
+    description: Object type returned by the API.
+    example: "video"
+  created_at:
+    type: integer
+    description: Unix timestamp (in seconds) of when the job was created.
+    example: 1777003604
+  model:
+    type: string
+    description: The model used for generation.
+    example: "wan2.2-t2v-a14b"
+  status:
+    type: string
+    description: >
+      Job status. Poll until the value is a terminal state such as `completed` or `failed`.
+    example: "processing"
+  output:
+    nullable: true
+    allOf:
+      - $ref: "video_generation_output.yml"
+  error:
+    type: object
+    nullable: true
+    description: Error details when status is `failed`.
+    additionalProperties: true
+  x_request_id:
+    type: string
+    nullable: true
+    description: Request correlation identifier when provided by the API.
diff --git a/specification/resources/inference/models/video_generation_output.yml b/specification/resources/inference/models/video_generation_output.yml
new file mode 100644
index 000000000..ecf51160d
--- /dev/null
+++ b/specification/resources/inference/models/video_generation_output.yml
@@ -0,0 +1,8 @@
+type: object
+description: Download location for a completed video when a presigned URL is available.
+properties:
+  url:
+    type: string
+    format: uri
+    description: Presigned URL for the generated MP4.
+    example: "https://example.com/output.mp4"