Skip to content

feat: add avalanche stealth provider #1705

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
steebchen merged 6 commits into from
Feb 20, 2026
Merged

feat: add avalanche stealth provider #1705

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
9094213
feat: add avalanche stealth provider
steebchen Feb 16, 2026
efbc22f
Merge remote-tracking branch 'origin/main' into add-avalanche-provider
steebchen Feb 20, 2026
b393e7b
feat: add models and vertex-style auth
steebchen Feb 20, 2026
88b8e11
Merge remote-tracking branch 'origin/main' into add-avalanche-provider
steebchen Feb 20, 2026
5f13481
fix: pass token to avalanche endpoint builder
steebchen Feb 20, 2026
8cbfd78
fix: avalanche auth, priority, and model specs
steebchen Feb 20, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
23 changes: 16 additions & 7 deletions apps/gateway/src/chat/chat.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1503,7 +1503,9 @@ chat.openapi(completions, async (c) => {
usedProvider,
providerKey?.baseUrl || undefined,
usedModel,
usedProvider === "google-ai-studio" || usedProvider === "google-vertex"
usedProvider === "google-ai-studio" ||
usedProvider === "google-vertex" ||
usedProvider === "avalanche"
? usedToken
: undefined,
stream,
Expand Down Expand Up @@ -2009,7 +2011,8 @@ chat.openapi(completions, async (c) => {
if (
usedProvider === "google-ai-studio" ||
usedProvider === "google-vertex" ||
usedProvider === "obsidian"
usedProvider === "obsidian" ||
usedProvider === "avalanche"
) {
const { redisClient } = await import("@llmgateway/cache");
for (const message of messages) {
Expand Down Expand Up @@ -3851,7 +3854,8 @@ chat.openapi(completions, async (c) => {
if (
usedProvider === "google-ai-studio" ||
usedProvider === "google-vertex" ||
usedProvider === "obsidian"
usedProvider === "obsidian" ||
usedProvider === "avalanche"
) {
const parts = data.candidates?.[0]?.content?.parts || [];
for (const part of parts) {
Expand All @@ -3878,7 +3882,8 @@ chat.openapi(completions, async (c) => {
} else if (
usedProvider === "google-ai-studio" ||
usedProvider === "google-vertex" ||
usedProvider === "obsidian"
usedProvider === "obsidian" ||
usedProvider === "avalanche"
) {
// For Google, count when grounding metadata is present
if (data.candidates?.[0]?.groundingMetadata) {
Expand Down Expand Up @@ -3974,6 +3979,7 @@ chat.openapi(completions, async (c) => {
case "google-ai-studio":
case "google-vertex":
case "obsidian":
case "avalanche":
// Preserve original Google finish reason for logging
if (data.promptFeedback?.blockReason) {
finishReason = data.promptFeedback.blockReason;
Expand Down Expand Up @@ -4372,7 +4378,8 @@ chat.openapi(completions, async (c) => {
const providerExcludesImageInput =
usedProvider === "google-ai-studio" ||
usedProvider === "google-vertex" ||
usedProvider === "obsidian";
usedProvider === "obsidian" ||
usedProvider === "avalanche";
const imageInputAdj = providerExcludesImageInput
? inputImageCount * 560
: 0;
Expand Down Expand Up @@ -5806,7 +5813,8 @@ chat.openapi(completions, async (c) => {
if (
usedProvider === "google-ai-studio" ||
usedProvider === "google-vertex" ||
usedProvider === "obsidian"
usedProvider === "obsidian" ||
usedProvider === "avalanche"
) {
logger.debug("Google model response parsed", {
usedProvider,
Expand Down Expand Up @@ -5984,7 +5992,8 @@ chat.openapi(completions, async (c) => {
const isGoogleContentFilter =
(usedProvider === "google-ai-studio" ||
usedProvider === "google-vertex" ||
usedProvider === "obsidian") &&
usedProvider === "obsidian" ||
usedProvider === "avalanche") &&
(finishReason === "SAFETY" ||
finishReason === "PROHIBITED_CONTENT" ||
finishReason === "RECITATION" ||
Expand Down
3 changes: 2 additions & 1 deletion apps/gateway/src/chat/tools/extract-content.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -7,7 +7,8 @@ export function extractContent(data: any, provider: Provider): string {
switch (provider) {
case "google-ai-studio":
case "google-vertex":
case "obsidian": {
case "obsidian":
case "avalanche": {
const parts = data.candidates?.[0]?.content?.parts || [];
const contentParts = parts.filter((part: any) => !part.thought);
return contentParts.map((part: any) => part.text).join("") || "";
Expand Down
3 changes: 2 additions & 1 deletion apps/gateway/src/chat/tools/extract-images.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -12,7 +12,8 @@ export function extractImages(data: any, provider: Provider): ImageObject[] {
switch (provider) {
case "google-ai-studio":
case "google-vertex":
case "obsidian": {
case "obsidian":
case "avalanche": {
const parts = data.candidates?.[0]?.content?.parts || [];
const imageParts = parts.filter((part: any) => part.inlineData);
return imageParts.map(
Expand Down
3 changes: 2 additions & 1 deletion apps/gateway/src/chat/tools/extract-reasoning.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -19,7 +19,8 @@ export function extractReasoning(data: any, provider: Provider): string {
}
case "google-ai-studio":
case "google-vertex":
case "obsidian": {
case "obsidian":
case "avalanche": {
const parts = data.candidates?.[0]?.content?.parts || [];
const reasoningParts = parts.filter((part: any) => part.thought);
return reasoningParts.map((part: any) => part.text).join("") || "";
Expand Down
1 change: 1 addition & 0 deletions apps/gateway/src/chat/tools/extract-token-usage.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,6 +50,7 @@ export function extractTokenUsage(
case "google-ai-studio":
case "google-vertex":
case "obsidian":
case "avalanche":
if (data.usageMetadata) {
promptTokens = data.usageMetadata.promptTokenCount ?? null;
let rawCandidates = data.usageMetadata.candidatesTokenCount ?? null;
Expand Down
3 changes: 2 additions & 1 deletion apps/gateway/src/chat/tools/extract-tool-calls.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,8 @@ export function extractToolCalls(data: any, provider: Provider): any[] | null {
return null;
case "google-ai-studio":
case "google-vertex":
case "obsidian": {
case "obsidian":
case "avalanche": {
// Google AI Studio tool calls in streaming
// Include thoughtSignature if present (required for Gemini 3 multi-turn conversations)
// Note: Redis caching of thought_signature happens in transform-streaming-to-openai.ts
Expand Down
3 changes: 2 additions & 1 deletion apps/gateway/src/chat/tools/parse-provider-response.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -181,7 +181,8 @@ export function parseProviderResponse(
}
case "google-ai-studio":
case "google-vertex":
case "obsidian": {
case "obsidian":
case "avalanche": {
// Check if response is missing candidates - treat as content filter
if (!json.candidates || json.candidates.length === 0) {
// Only log warning if there's no blockReason explaining why
Expand Down
3 changes: 2 additions & 1 deletion apps/gateway/src/chat/tools/transform-response-to-openai.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -87,7 +87,8 @@ export function transformResponseToOpenai(
switch (usedProvider) {
case "google-ai-studio":
case "google-vertex":
case "obsidian": {
case "obsidian":
case "avalanche": {
transformedResponse = {
id: `chatcmpl-${Date.now()}`,
object: "chat.completion",
Expand Down
3 changes: 2 additions & 1 deletion apps/gateway/src/chat/tools/transform-streaming-to-openai.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -251,7 +251,8 @@ export function transformStreamingToOpenai(

case "google-ai-studio":
case "google-vertex":
case "obsidian": {
case "obsidian":
case "avalanche": {
const mapFinishReason = (
finishReason?: string,
hasFunctionCalls?: boolean,
Expand Down
6 changes: 4 additions & 2 deletions apps/gateway/src/lib/costs.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -312,7 +312,8 @@ export async function calculateCosts(
const isGoogleProvider =
provider === "google-ai-studio" ||
provider === "google-vertex" ||
provider === "obsidian";
provider === "obsidian" ||
provider === "avalanche";
const totalOutputTokens = isGoogleProvider
? calculatedCompletionTokens
: calculatedCompletionTokens + (reasoningTokens || 0);
Expand Down Expand Up @@ -384,7 +385,8 @@ export async function calculateCosts(
imageInputTokens &&
(provider === "google-ai-studio" ||
provider === "google-vertex" ||
provider === "obsidian")
provider === "obsidian" ||
provider === "avalanche")
? (calculatedPromptTokens || 0) + imageInputTokens
: calculatedPromptTokens,
completionTokens: calculatedCompletionTokens,
Expand Down
4 changes: 3 additions & 1 deletion apps/gateway/src/lib/logs.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -21,7 +21,8 @@ export function isExpectedUnknownFinishReason(
if (
(provider === "google-ai-studio" ||
provider === "google-vertex" ||
provider === "obsidian") &&
provider === "obsidian" ||
provider === "avalanche") &&
finishReason === "OTHER"
) {
return true;
Expand Down Expand Up @@ -74,6 +75,7 @@ export function getUnifiedFinishReason(
case "google-ai-studio":
case "google-vertex":
case "obsidian":
case "avalanche":
// Google finish reasons (original format, not mapped to OpenAI)
if (finishReason === "STOP") {
return UnifiedFinishReason.COMPLETED;
Expand Down
40 changes: 40 additions & 0 deletions packages/actions/src/get-provider-endpoint.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -69,6 +69,14 @@ export function getProviderEndpoint(
);
}
break;
case "avalanche":
url = getProviderEnvValue("avalanche", "baseUrl", configIndex);
if (!url) {
throw new Error(
"Avalanche provider requires LLM_AVALANCHE_BASE_URL environment variable",
);
}
break;
case "inference.net":
url = "https://api.inference.net";
break;
Expand Down Expand Up @@ -198,6 +206,38 @@ export function getProviderEndpoint(
? `${baseEndpoint}?${queryParams.join("&")}`
: baseEndpoint;
}
case "avalanche": {
const endpoint = stream ? "streamGenerateContent" : "generateContent";
const model = modelName || "gemini-3-pro-image-preview";

const projectId = getProviderEnvValue(
"avalanche",
"project",
configIndex,
);

const region =
getProviderEnvValue("avalanche", "region", configIndex, "global") ||
"global";

if (!projectId) {
throw new Error(
"Avalanche provider requires LLM_AVALANCHE_PROJECT environment variable",
);
}

const baseEndpoint = `${url}/v1/projects/${projectId}/locations/${region}/publishers/google/models/${model}:${endpoint}`;
const queryParams = [];
if (token) {
queryParams.push(`key=${token}`);
}
if (stream) {
queryParams.push("alt=sse");
}
return queryParams.length > 0
? `${baseEndpoint}?${queryParams.join("&")}`
: baseEndpoint;
}
case "google-vertex": {
const endpoint = stream ? "streamGenerateContent" : "generateContent";
const model = modelName || "gemini-2.5-flash-lite";
Expand Down
1 change: 1 addition & 0 deletions packages/actions/src/get-provider-headers.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -29,6 +29,7 @@ export function getProviderHeaders(
}
case "google-ai-studio":
case "google-vertex":
case "avalanche":
return {};
case "obsidian":
return {
Expand Down
3 changes: 2 additions & 1 deletion packages/actions/src/prepare-request-body.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1340,7 +1340,8 @@ export async function prepareRequestBody(
}
case "google-ai-studio":
case "google-vertex":
case "obsidian": {
case "obsidian":
case "avalanche": {
delete requestBody.model; // Not used in body
delete requestBody.stream; // Stream is handled via URL parameter
delete requestBody.messages; // Not used in body for Google providers
Expand Down
Loading
Loading