@aliou/pi-neuralwatt 0.14.2 → 0.15.0

package/extensions/provider/models/early-access.ts

@@ -1,158 +0,0 @@
-import type { ProviderModelConfig } from "@earendil-works/pi-coding-agent";
-import { fetchNeuralwattModels } from "../../../src/lib/neuralwatt-api";
-import type { NeuralwattApiModel } from "../../../src/types/models-api";
-import { buildThinkingLevelMap, resolveMaxTokens } from "./build";
-import { NEURALWATT_MODELS } from "./public-models";
-
-// Pre-release models. Neuralwatt ships these to authorized accounts before they
-// reach the public /v1/models response; most go public eventually. Keep them
-// gated by includeEarlyAccessModels and hardcode entries so they remain
-// available from the offline catalog.
-// Move an entry to public-models.ts once Neuralwatt advertises it publicly.
-export const EARLY_ACCESS_NEURALWATT_MODELS: ProviderModelConfig[] = [
-  // Qwen3.8 27B FP8 — pre-release dense 27B VL model with MTP speculative
-  // decoding, native 262K context, served on 2x H200. Returned by the
-  // authenticated /v1/models catalog but absent from the public endpoint.
-  // Binary thinking is toggled through `chat_template_kwargs.enable_thinking`
-  // (verified: `enable_thinking: false` suppresses reasoning output); the API
-  // exposes no `reasoning` block, so the thinking level map is the high-only
-  // fallback with `off: null`.
-  // https://huggingface.co/Qwen/Qwen3.8-27B-FP8
-  {
-    id: "Qwen/Qwen3.8-27B-FP8",
-    name: "Qwen 3.8 27B FP8",
-    reasoning: true,
-    input: ["text", "image"],
-    cost: { input: 0.45, output: 3.2, cacheRead: 0.25, cacheWrite: 0 },
-    contextWindow: 262_128,
-    maxTokens: 65_536,
-    compat: {
-      supportsDeveloperRole: false,
-      maxTokensField: "max_tokens",
-      requiresReasoningContentOnAssistantMessages: true,
-      thinkingFormat: "chat-template",
-      chatTemplateKwargs: {
-        enable_thinking: { $var: "thinking.enabled" },
-      },
-    },
-    thinkingLevelMap: { ...buildThinkingLevelMap(undefined) },
-  },
-];
-
-// Per-ID overrides for known early-access models. The authenticated /v1/models
-// endpoint exposes pricing and capabilities, but some Pi-specific behavior
-// (compat flags, context window, max tokens) has to be supplied by hand.
-// Reasoning config is always derived from the endpoint's `reasoning` block via
-// `buildThinkingLevelMap`, so `reasoning` and `thinkingLevelMap` are not part
-// of the override shape.
-// Models that have since gone public now live in public-models.ts.
-const EARLY_ACCESS_MODEL_OVERRIDES: Partial<
-  Record<
-    string,
-    Omit<Partial<ProviderModelConfig>, "reasoning" | "thinkingLevelMap">
-  >
-> = {};
-
-function buildEarlyAccessModel(
-  apiModel: NeuralwattApiModel,
-): ProviderModelConfig {
-  const meta = apiModel.metadata;
-  const reasoning = meta?.capabilities.reasoning ?? false;
-  const override = EARLY_ACCESS_MODEL_OVERRIDES[apiModel.id];
-
-  const compat: NonNullable<ProviderModelConfig["compat"]> = {
-    supportsDeveloperRole: false,
-    maxTokensField: "max_tokens",
-  };
-  if (reasoning) {
-    compat.requiresReasoningContentOnAssistantMessages = true;
-  }
-
-  const model: ProviderModelConfig = {
-    id: apiModel.id,
-    name: meta?.display_name ?? apiModel.id,
-    reasoning,
-    input: (meta?.capabilities.vision ? ["text", "image"] : ["text"]) as (
-      | "text"
-      | "image"
-    )[],
-    cost: {
-      input: meta?.pricing.input_per_million ?? 0,
-      output: meta?.pricing.output_per_million ?? 0,
-      cacheRead: meta?.pricing.cached_input_per_million ?? 0,
-      cacheWrite: meta?.pricing.cached_output_per_million ?? 0,
-    },
-    contextWindow: apiModel.max_model_len,
-    maxTokens: resolveMaxTokens(
-      meta?.limits.max_output_tokens,
-      apiModel.max_model_len,
-    ),
-    compat,
-  };
-
-  if (reasoning) {
-    // Reasoning levels come straight from the endpoint's `reasoning` block:
-    // `supported_efforts` (identity) and `mandatory` (gates `off`). A missing
-    // block falls back to a high-only map inside `buildThinkingLevelMap`.
-    model.thinkingLevelMap = buildThinkingLevelMap(meta?.reasoning);
-  }
-
-  if (override) {
-    return applyEarlyAccessOverride(model, override);
-  }
-
-  return model;
-}
-
-function applyEarlyAccessOverride(
-  model: ProviderModelConfig,
-  override: Partial<ProviderModelConfig>,
-): ProviderModelConfig {
-  const result: ProviderModelConfig = { ...model };
-
-  if (override.name !== undefined) result.name = override.name;
-  if (override.input !== undefined) result.input = override.input;
-  if (override.contextWindow !== undefined) {
-    result.contextWindow = override.contextWindow;
-  }
-  if (override.maxTokens !== undefined) result.maxTokens = override.maxTokens;
-  if (override.cost !== undefined) {
-    result.cost = { ...model.cost, ...override.cost };
-  }
-  if (override.compat !== undefined) {
-    result.compat = { ...model.compat, ...override.compat };
-  }
-
-  // `reasoning` and `thinkingLevelMap` are intentionally not overridable:
-  // reasoning config is derived from the endpoint's `reasoning` block.
-
-  return result;
-}
-
-/**
- * Load early-access models from the authenticated /v1/models endpoint.
- *
- * Early-access models are any models returned by the API that are not already
- * part of the public hardcoded list. If the API key is missing or the request
- * fails, an `undefined` distinguishes an unavailable/failed request from a
- * successful empty list, allowing refresh callers to preserve stale cache.
- */
-export async function loadEarlyAccessModels(
-  apiKey: string,
-  signal?: AbortSignal,
-): Promise<ProviderModelConfig[] | undefined> {
-  if (!apiKey) return undefined;
-
-  const result = await fetchNeuralwattModels(apiKey, signal);
-  if (!result.success) return undefined;
-
-  const publicIds = new Set(NEURALWATT_MODELS.map((model) => model.id));
-
-  return result.data
-    .filter(
-      (model) =>
-        !model.metadata?.deprecated && !model.metadata?.pricing.pricing_tbd,
-    )
-    .filter((model) => !publicIds.has(model.id))
-    .map(buildEarlyAccessModel);
-}

package/extensions/provider/models/legacy.ts

@@ -1,47 +0,0 @@
-import type { ProviderModelConfig } from "@earendil-works/pi-coding-agent";
-import { NEURALWATT_MODELS } from "./public-models";
-
-// Legacy model IDs that should resolve to a canonical public model.
-// These are phased out over time and are only included when `includeLegacyModelIds` is enabled.
-export const LEGACY_MODEL_ALIAS_MAP = {
-  // GLM 5.1 → GLM 5.2
-  "glm-5.1": "glm-5.2",
-  "glm-5.1-fast": "glm-5.2-fast",
-  "zai-org/GLM-5.1-FP8": "glm-5.2",
-  // Kimi K2.5 → Kimi K2.7 Code (K2.6 was retired 8/3, redirected to K2.7)
-  "moonshotai/Kimi-K2.5": "kimi-k2.7-code",
-  "kimi-k2.5-fast": "kimi-k2.7-code-fast",
-  // Kimi K2.6 retired 8/3, redirected to Kimi K2.7 Code
-  "kimi-k2.6": "kimi-k2.7-code",
-  "kimi-k2.6-fast": "kimi-k2.7-code-fast",
-  "kimi-k2.6-flex": "kimi-k2.7-code-flex",
-  "moonshotai/Kimi-K2.6": "kimi-k2.7-code",
-  // Qwen 3.5 retired 8/3, redirected to Qwen 3.6
-  "qwen3.5-397b": "qwen3.6-35b",
-  "qwen3.5-397b-fast": "qwen3.6-35b-fast",
-  "Qwen/Qwen3.5-397B-A17B-FP8": "qwen3.6-35b",
-} as const;
-
-export const LEGACY_NEURALWATT_MODEL_IDS = new Set<string>(
-  Object.keys(LEGACY_MODEL_ALIAS_MAP),
-);
-
-export function buildLegacyNeuralwattModels(): ProviderModelConfig[] {
-  return Object.entries(LEGACY_MODEL_ALIAS_MAP).map(
-    ([legacyId, canonicalId]) => {
-      const canonical = NEURALWATT_MODELS.find(
-        (model) => model.id === canonicalId,
-      );
-
-      if (!canonical) {
-        throw new Error(`Missing canonical model for legacy alias ${legacyId}`);
-      }
-
-      return {
-        ...canonical,
-        id: legacyId,
-        name: `${canonical.name} (legacy ID)`,
-      };
-    },
-  );
-}