Never stop coding. Free MIT AI gateway: one endpoint, 375 providers (150+ free), 1,400+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 900+ contributors
TypeScript a2aai-agentsai-gatewayanthropicclaudeclaude-codeclinecodexcopilotcursordeepseekfree-aigeminikimillm-gatewaymcpopenaiopenai-proxyqwentoken-saver
import assert from "node:assert/strict";
import fs from "node:fs";
import os from "node:os";
import path from "node:path";
import test from "node:test";
// Copilot ships Responses-only GPT ids (gpt-6-luna, gpt-6.1-sol, gpt-5.5, ...) before the
// static seed knows them and before a live /models sync has run. Without a format hint the
// request goes out as a Chat Completions body to /chat/completions and Copilot answers 400
// ("... use /v1/responses"). Explicit sources (custom > compat > synced > static) must still win.
const dataDir = fs.mkdtempSync(path.join(os.tmpdir(), "omniroute-copilot-unsynced-"));
process.env.DATA_DIR = dataDir;
process.env.API_KEY_SECRET ||= "copilot-unsynced-test-secret";
const core = await import("../../src/lib/db/core.ts");
const providers = await import("../../src/lib/db/providers.ts");
const modelsDb = await import("../../src/lib/db/models.ts");
const { persistDiscoveredModels } = await import("../../src/lib/providerModels/modelDiscovery.ts");
const { getModelInfo } = await import("../../src/sse/services/model.ts");
const { parseGitHubCopilotModels } = await import("../../open-sse/services/githubCopilotModels.ts");
const { resolveChatCoreTargetFormat } =
await import("../../open-sse/handlers/chatCore/targetFormat.ts");
const { resolveExecutionCredentials } =
await import("../../open-sse/handlers/chatCore/executionCredentials.ts");
const { GithubExecutor } = await import("../../open-sse/executors/github.ts");
const RESPONSES_URL = "https://api.githubcopilot.com/responses";
const CHAT_URL = "https://api.githubcopilot.com/chat/completions";
const executor = new GithubExecutor();
test.after(() => {
core.resetDbInstance();
fs.rmSync(dataDir, { recursive: true, force: true, maxRetries: 5, retryDelay: 100 });
});
async function resolveRouting(model: string) {
const info = await getModelInfo(`gh/${model}`);
assert.equal(info.provider, "github");
assert.equal(info.model, model);
const { targetFormat } = resolveChatCoreTargetFormat({
provider: info.provider,
resolvedModel: info.model,
apiFormat: info.apiFormat,
sourceFormat: "openai",
customModelTargetFormat: info.targetFormat,
providerSpecificData: {},
});
const credentials = resolveExecutionCredentials({
credentials: { providerSpecificData: {} },
provider: info.provider,
targetFormat,
nativeCodexPassthrough: false,
endpointPath: "/v1/chat/completions",
ccSessionId: null,
});
return {
infoTargetFormat: info.targetFormat,
targetFormat,
url: executor.buildUrl(info.model, false, 0, credentials),
};
}
test("unsynced Responses-only GPT ids route to /responses with a Responses body", async () => {
for (const model of ["gpt-6-luna", "gpt-6-sol", "gpt-6.1-sol", "gpt-5.5", "gpt-7-nova"]) {
assert.deepEqual(
await resolveRouting(model),
{
infoTargetFormat: "openai-responses",
targetFormat: "openai-responses",
url: RESPONSES_URL,
},
model
);
}
});
test("older GPT ids outside the heuristic keep Chat Completions", async () => {
for (const model of ["gpt-5.4-mini", "gpt-5.1", "gpt-4.1"]) {
const routing = await resolveRouting(model);
assert.equal(routing.infoTargetFormat, undefined, model);
assert.equal(routing.targetFormat, "openai", model);
assert.equal(routing.url, CHAT_URL, model);
}
});
test("Claude and Gemini ids are never inferred as openai-responses", async () => {
for (const model of ["claude-opus-5", "claude-gpt-6-bridge"]) {
const routing = await resolveRouting(model);
assert.notEqual(routing.infoTargetFormat, "openai-responses", model);
assert.equal(routing.targetFormat, "claude", model);
assert.equal(routing.url, "https://api.githubcopilot.com/v1/messages", model);
}
for (const model of ["gemini-3.8-flash", "gpt-6-gemini-mix"]) {
const routing = await resolveRouting(model);
assert.notEqual(routing.infoTargetFormat, "openai-responses", model);
assert.equal(routing.url, CHAT_URL, model);
}
});
test("the heuristic is scoped to github and does not touch other providers", async () => {
const info = await getModelInfo("openrouter/gpt-6-luna");
assert.notEqual(info.targetFormat, "openai-responses");
});
test("an explicit custom-model targetFormat wins over the heuristic", async () => {
await modelsDb.addCustomModel(
"github",
"gpt-6-custom-chat",
"GPT-6 custom chat",
"manual",
"chat-completions",
["chat"],
"openai"
);
const routing = await resolveRouting("gpt-6-custom-chat");
assert.equal(routing.infoTargetFormat, "openai");
assert.equal(routing.targetFormat, "openai");
assert.equal(routing.url, CHAT_URL);
});
test("a synced chat-capable row wins over the heuristic", async () => {
const connection = await providers.createProviderConnection({
provider: "github",
authType: "oauth",
name: "unsynced-heuristic-test",
isActive: true,
testStatus: "active",
});
const rows = parseGitHubCopilotModels({
data: [
{ id: "gpt-6-luna", supported_endpoints: ["/chat/completions"] },
{ id: "gpt-6.1-sol", supported_endpoints: ["/responses"] },
],
});
await persistDiscoveredModels("github", connection.id, rows);
const chatRow = await resolveRouting("gpt-6-luna");
assert.equal(chatRow.infoTargetFormat, undefined);
assert.equal(chatRow.targetFormat, "openai");
assert.equal(chatRow.url, CHAT_URL);
assert.deepEqual(await resolveRouting("gpt-6.1-sol"), {
infoTargetFormat: "openai-responses",
targetFormat: "openai-responses",
url: RESPONSES_URL,
});
});
test("GithubExecutor.buildUrl sends gpt-6-luna to /responses once the format is resolved", () => {
const credentials = resolveExecutionCredentials({
credentials: { providerSpecificData: {} },
provider: "github",
targetFormat: "openai-responses",
nativeCodexPassthrough: false,
endpointPath: "/v1/chat/completions",
ccSessionId: null,
});
assert.equal(executor.buildUrl("gpt-6-luna", false, 0, credentials), RESPONSES_URL);
});