pi-wafer-provider 1.0.2

package/models.json

@@ -0,0 +1,220 @@
+[
+  {
+    "id": "deepseek-v4-flash",
+    "name": "Deepseek V4 Flash",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 1000000,
+    "maxTokens": 16384,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "supportsZdr": true
+    }
+  },
+  {
+    "id": "deepseek-v4-pro",
+    "name": "Deepseek V4 Pro",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 1000000,
+    "maxTokens": 16384,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "supportsZdr": true
+    }
+  },
+  {
+    "id": "GLM-5.1",
+    "name": "GLM 5.1",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 1.5,
+      "output": 4.5,
+      "cacheRead": 0.15,
+      "cacheWrite": 0
+    },
+    "contextWindow": 202752,
+    "maxTokens": 32768,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsZdr": true
+    }
+  },
+  {
+    "id": "GLM-5.2",
+    "name": "GLM 5.2",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 1048576,
+    "maxTokens": 16384,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "supportsZdr": true
+    }
+  },
+  {
+    "id": "Kimi-K2.6",
+    "name": "Kimi K2.6",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 1.1,
+      "output": 4.8,
+      "cacheRead": 0.11,
+      "cacheWrite": 0
+    },
+    "contextWindow": 262144,
+    "maxTokens": 32768,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsZdr": false
+    }
+  },
+  {
+    "id": "Kimi-K2.7-Code",
+    "name": "Kimi K2.7 Code",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 262144,
+    "maxTokens": 16384,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "supportsZdr": false
+    }
+  },
+  {
+    "id": "MiniMax-M3",
+    "name": "MiniMax M3",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 1048576,
+    "maxTokens": 16384,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "supportsZdr": false
+    }
+  },
+  {
+    "id": "Qwen3.5-397B-A17B",
+    "name": "Qwen 3.5 397B (A17B)",
+    "reasoning": false,
+    "input": [
+      "text",
+      "image"
+    ],
+    "cost": {
+      "input": 0.6,
+      "output": 3.6,
+      "cacheRead": 0.06,
+      "cacheWrite": 0
+    },
+    "contextWindow": 262144,
+    "maxTokens": 32768,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsZdr": false
+    }
+  },
+  {
+    "id": "Qwen3.6-35B-A3B",
+    "name": "Qwen3.6 35B A3B",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 256000,
+    "maxTokens": 16384,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "supportsZdr": false
+    }
+  },
+  {
+    "id": "qwen3.7-max",
+    "name": "Qwen3.7 Max",
+    "reasoning": false,
+    "input": [
+      "text"
+    ],
+    "cost": {
+      "input": 0,
+      "output": 0,
+      "cacheRead": 0,
+      "cacheWrite": 0
+    },
+    "contextWindow": 256000,
+    "maxTokens": 16384,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsStore": false,
+      "supportsZdr": false
+    }
+  }
+]

package/package.json

@@ -0,0 +1,31 @@
+{
+  "name": "pi-wafer-provider",
+  "version": "1.0.2",
+  "description": "Wafer Serverless provider extension for pi - Access Qwen3.5-397B-A17B, GLM-5.1, Kimi K2.6, and DeepSeek-V4 through the Wafer Serverless API",
+  "type": "module",
+  "main": "index.ts",
+  "keywords": [
+    "pi",
+    "extension",
+    "provider",
+    "wafer",
+    "wafer-serverless",
+    "ai",
+    "llm",
+    "qwen",
+    "glm"
+  ],
+  "author": "",
+  "license": "MIT",
+  "pi": {
+    "extensions": [
+      "./index.ts"
+    ]
+  },
+  "scripts": {
+    "clean": "echo 'nothing to clean'",
+    "build": "echo 'nothing to build'",
+    "check": "echo 'nothing to check'",
+    "update-models": "node scripts/update-models.js"
+  }
+}

package/patch.json

@@ -0,0 +1,86 @@
+{
+  "Qwen3.5-397B-A17B": {
+    "providers": ["wafer-serverless"],
+    "reasoning": true,
+    "compat": {
+      "thinkingFormat": "qwen",
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsReasoningEffort": true
+    }
+  },
+  "GLM-5.1": {
+    "providers": ["wafer-serverless"],
+    "reasoning": true,
+    "compat": {
+      "thinkingFormat": "zai",
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsReasoningEffort": true
+    }
+  },
+  "Kimi-K2.6": {
+    "providers": ["wafer-serverless"],
+    "reasoning": true,
+    "compat": {
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsReasoningEffort": true
+    }
+  },
+  "Qwen3.6-35B-A3B": {
+    "providers": ["wafer-serverless"],
+    "reasoning": true,
+    "compat": {
+      "thinkingFormat": "qwen",
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsReasoningEffort": true
+    }
+  },
+  "deepseek-v4-flash": {
+    "providers": ["wafer-serverless"],
+    "reasoning": true,
+    "compat": {
+      "thinkingFormat": "deepseek",
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsReasoningEffort": true
+    }
+  },
+  "deepseek-v4-pro": {
+    "providers": ["wafer-serverless"],
+    "reasoning": true,
+    "contextWindow": 1000000,
+    "maxTokens": 384000,
+    "thinkingLevelMap": {
+      "minimal": null,
+      "low": null,
+      "medium": null,
+      "high": "high",
+      "xhigh": "max"
+    },
+    "compat": {
+      "thinkingFormat": "deepseek",
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsReasoningEffort": true
+    }
+  },
+  "qwen3.7-max": {
+    "providers": ["wafer-serverless"],
+    "reasoning": true,
+    "cost": {
+      "input": 5.0,
+      "output": 15.0,
+      "cacheRead": 0.5,
+      "cacheWrite": 0
+    },
+    "compat": {
+      "thinkingFormat": "qwen",
+      "maxTokensField": "max_completion_tokens",
+      "supportsDeveloperRole": false,
+      "supportsReasoningEffort": true
+    }
+  }
+}

package/scripts/e2e-kimi-k2.6.mjs

@@ -0,0 +1,218 @@
+#!/usr/bin/env node
+/**
+ * E2E test for Kimi K2.6 on Wafer (pass.wafer.ai).
+ *
+ * Tests:
+ * 1. Basic streaming with reasoning prompt (delta.reasoning / reasoning_content)
+ * 2. Non-streaming (reasoning_content structure)
+ * 3. Preserved Thinking via thinking.keep="all"
+ * 4. Preserved Thinking via chat_template_kwargs
+ */
+
+const API_KEY = process.env.WAFER_SERVERLESS_API_KEY || process.env.WAFER_API_KEY;
+const MODEL = process.env.MODEL || "Kimi-K2.6";
+const BASE_URL = "https://pass.wafer.ai/v1/chat/completions";
+
+if (!API_KEY) {
+  console.error("❌ WAFER_SERVERLESS_API_KEY (or WAFER_API_KEY) is not set.");
+  process.exit(1);
+}
+
+async function streamRequest(label, body) {
+  console.log(`\n═══════════════════════════════════════════════════`);
+  console.log(`  TEST: ${label}`);
+  console.log(`  Model: ${MODEL}`);
+  console.log(`  Base URL: ${BASE_URL}`);
+  console.log(`═══════════════════════════════════════════════════`);
+
+  const res = await fetch(BASE_URL, {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      Authorization: `Bearer ${API_KEY}`,
+      "Wafer-ZDR": "required",
+    },
+    body: JSON.stringify(body),
+  });
+
+  if (!res.ok) {
+    const text = await res.text();
+    console.error(`❌ HTTP ${res.status}: ${text}`);
+    return { ok: false, error: text };
+  }
+
+  const reader = res.body.getReader();
+  const decoder = new TextDecoder();
+  let buffer = "";
+
+  let reasoningChunks = [];
+  let contentChunks = [];
+  let hasDone = false;
+
+  try {
+    while (true) {
+      const { done, value } = await reader.read();
+      if (done) break;
+
+      buffer += decoder.decode(value, { stream: true });
+      const lines = buffer.split("\n");
+      buffer = lines.pop() || "";
+
+      for (const line of lines) {
+        const trimmed = line.trim();
+        if (trimmed.startsWith("data: ")) {
+          const payload = trimmed.slice(6);
+          if (payload === "[DONE]") {
+            hasDone = true;
+            continue;
+          }
+          try {
+            const obj = JSON.parse(payload);
+            const delta = obj.choices?.[0]?.delta || {};
+            if (delta.reasoning) reasoningChunks.push(delta.reasoning);
+            if (delta.content) contentChunks.push(delta.content);
+          } catch {}
+        }
+      }
+    }
+  } catch (e) {
+    console.error(`💥 Stream error: ${e.message}`);
+  } finally {
+    reader.releaseLock();
+  }
+
+  console.log(`  Reasoning chunks:  ${reasoningChunks.length}`);
+  console.log(`  Content chunks:      ${contentChunks.length}`);
+  console.log(`  [DONE] seen:       ${hasDone}`);
+
+  if (reasoningChunks.length > 0) {
+    console.log(`\n  --- First reasoning chunk ---`);
+    console.log(`  ${reasoningChunks[0].substring(0, 200)}`);
+  }
+  if (contentChunks.length > 0) {
+    const fullContent = contentChunks.join("");
+    console.log(`\n  --- Final content (first 300 chars) ---`);
+    console.log(`  ${fullContent.substring(0, 300)}`);
+  }
+
+  return { ok: true, reasoning: reasoningChunks, content: contentChunks };
+}
+
+async function nonStreamRequest(label, body) {
+  console.log(`\n═══════════════════════════════════════════════════`);
+  console.log(`  TEST: ${label}`);
+  console.log(`  Model: ${MODEL}`);
+  console.log(`  Base URL: ${BASE_URL}`);
+  console.log(`═══════════════════════════════════════════════════`);
+
+  const res = await fetch(BASE_URL, {
+    method: "POST",
+    headers: {
+      "Content-Type": "application/json",
+      Authorization: `Bearer ${API_KEY}`,
+      "Wafer-ZDR": "required",
+    },
+    body: JSON.stringify(body),
+  });
+
+  const data = await res.json();
+
+  if (!res.ok) {
+    console.error(`❌ HTTP ${res.status}: ${JSON.stringify(data)}`);
+    return { ok: false, error: data };
+  }
+
+  const choice = data.choices?.[0];
+  const msg = choice?.message || {};
+
+  console.log(`  keys:                ${JSON.stringify(Object.keys(msg))}`);
+  console.log(`  has reasoning_content: ${"reasoning_content" in msg}`);
+  console.log(`  has reasoning:         ${"reasoning" in msg}`);
+  if (msg.reasoning_content) {
+    console.log(`  reasoning_content len: ${msg.reasoning_content.length}`);
+    console.log(`  --- reasoning_content (first 300) ---`);
+    console.log(`  ${msg.reasoning_content.substring(0, 300)}`);
+  }
+  if (msg.reasoning) {
+    console.log(`  reasoning: ${msg.reasoning.substring(0, 300)}`);
+  }
+  console.log(`  content: ${(msg.content || "").substring(0, 300)}`);
+  console.log(`  usage: ${JSON.stringify(data.usage)}`);
+
+  return { ok: true, message: msg, usage: data.usage };
+}
+
+async function main() {
+  console.log(`E2E Testing Kimi K2.6 on Wafer`);
+  console.log(`Base URL: ${BASE_URL}`);
+
+  // Test 1: Basic streaming
+  await streamRequest("Basic stream (reasoning prompt)", {
+    model: MODEL,
+    messages: [{ role: "user", content: "What is 13 × 17? Show your reasoning." }],
+    max_tokens: 500,
+    stream: true,
+  });
+
+  // Test 2: Non-streaming
+  await nonStreamRequest("Non-stream (inspect structure)", {
+    model: MODEL,
+    messages: [{ role: "user", content: "What is 13 × 17? Show your reasoning." }],
+    max_tokens: 500,
+    stream: false,
+  });
+
+  // Test 3: Preserved Thinking via extra_body thinking.keep
+  const r3 = await streamRequest("Stream with thinking.keep='all' (extra_body)", {
+    model: MODEL,
+    messages: [
+      { role: "user", content: "Tell me three random numbers." },
+      {
+        role: "assistant",
+        reasoning_content: "I generated five numbers and told the first three: 473, 921, 235, and the other two were 215, 222.",
+        content: "473, 921, 235",
+      },
+      { role: "user", content: "What are the other two numbers you have in mind?" },
+    ],
+    max_tokens: 500,
+    stream: true,
+    extra_body: { thinking: { type: "enabled", keep: "all" } },
+  });
+
+  if (r3.ok && r3.content.length > 0) {
+    const fullContent = r3.content.join("");
+    const got215 = fullContent.includes("215");
+    const got222 = fullContent.includes("222");
+    console.log(`\n  🧪 Preserved Thinking (extra_body) check: mentions 215=${got215}, mentions 222=${got222}`);
+  }
+
+  // Test 4: Preserved Thinking via chat_template_kwargs
+  const r4 = await streamRequest("Stream with chat_template_kwargs preserve_thinking", {
+    model: MODEL,
+    messages: [
+      { role: "user", content: "Tell me three random numbers." },
+      {
+        role: "assistant",
+        reasoning_content: "I generated five numbers and told the first three: 473, 921, 235, and kept 215 and 222 hidden.",
+        content: "473, 921, 235",
+      },
+      { role: "user", content: "What are the other two numbers you have in mind?" },
+    ],
+    max_tokens: 500,
+    stream: true,
+    extra_body: { chat_template_kwargs: { thinking: true, preserve_thinking: true } },
+  });
+
+  if (r4.ok && r4.content.length > 0) {
+    const fullContent = r4.content.join("");
+    const got215 = fullContent.includes("215");
+    const got222 = fullContent.includes("222");
+    console.log(`\n  🧪 Preserved Thinking (chat_template_kwargs) check: mentions 215=${got215}, mentions 222=${got222}`);
+  }
+
+  console.log(`\n═══════════════════════════════════════════════════`);
+  console.log(`  E2E Test Complete`);
+  console.log(`═══════════════════════════════════════════════════`);
+}
+
+main().catch((e) => { console.error(e); process.exit(1); });