summaryrefslogtreecommitdiffhomepage
diff options
context:
space:
mode:
Diffstat
-rw-r--r--cloud/app/src/routes/zen/v1/chat/completions.ts378
-rw-r--r--cloud/app/src/routes/zen/v1/responses.ts34
-rw-r--r--cloud/app/src/util/zen.ts367
-rw-r--r--cloud/function/sst-env.d.ts4
-rw-r--r--cloud/resource/resource.cloudflare.ts3
-rw-r--r--cloud/resource/sst-env.d.ts4
-rw-r--r--infra/cloud.ts1
-rw-r--r--packages/function/sst-env.d.ts4
-rw-r--r--packages/opencode/src/provider/models.ts1
-rw-r--r--packages/opencode/src/provider/provider.ts7
-rw-r--r--sst-env.d.ts4
11 files changed, 451 insertions, 356 deletions
diff --git a/cloud/app/src/routes/zen/v1/chat/completions.ts b/cloud/app/src/routes/zen/v1/chat/completions.ts
index cf0dbcabd..0a3b0e759 100644
--- a/cloud/app/src/routes/zen/v1/chat/completions.ts
+++ b/cloud/app/src/routes/zen/v1/chat/completions.ts
@@ -1,361 +1,33 @@
-import { Resource } from "@opencode/cloud-resource"
import type { APIEvent } from "@solidjs/start/server"
-import { Database, eq, sql } from "@opencode/cloud-core/drizzle/index.js"
-import { KeyTable } from "@opencode/cloud-core/schema/key.sql.js"
-import { BillingTable, UsageTable } from "@opencode/cloud-core/schema/billing.sql.js"
-import { centsToMicroCents } from "@opencode/cloud-core/util/price.js"
-import { Identifier } from "@opencode/cloud-core/identifier.js"
-
-const MODELS = {
- // "anthropic/claude-sonnet-4": {
- // auth: true,
- // api: "https://api.anthropic.com",
- // apiKey: Resource.ANTHROPIC_API_KEY.value,
- // model: "claude-sonnet-4-20250514",
- // cost: {
- // input: 0.0000015,
- // output: 0.000006,
- // reasoning: 0.0000015,
- // cacheRead: 0.0000001,
- // cacheWrite: 0.0000001,
- // },
- // headerMappings: {},
- // },
- "gpt-5": {
- id: "gpt-5" as const,
- auth: true,
- api: "https://inference.baseten.co",
- apiKey: Resource.BASETEN_API_KEY.value,
- model: "Qwen/Qwen3-Coder-480B-A35B-Instruct",
- cost: {
- input: 0.00000125,
- output: 0.00001,
- reasoning: 0,
- cacheRead: 0.000000125,
- cacheWrite: 0,
- },
- headerMappings: {},
- },
- "qwen3-coder": {
- id: "qwen3-coder" as const,
- auth: true,
- api: "https://inference.baseten.co",
- apiKey: Resource.BASETEN_API_KEY.value,
- model: "Qwen/Qwen3-Coder-480B-A35B-Instruct",
- cost: {
- input: 0.00000038,
- output: 0.00000153,
- reasoning: 0,
- cacheRead: 0,
- cacheWrite: 0,
- },
- headerMappings: {},
- },
- "kimi-k2": {
- id: "kimi-k2" as const,
- auth: true,
- api: "https://inference.baseten.co",
- apiKey: Resource.BASETEN_API_KEY.value,
- model: "moonshotai/Kimi-K2-Instruct-0905",
- cost: {
- input: 0.0000006,
- output: 0.0000025,
- reasoning: 0,
- cacheRead: 0,
- cacheWrite: 0,
- },
- headerMappings: {},
- },
- "grok-code": {
- id: "grok-code" as const,
- auth: false,
- api: "https://api.x.ai",
- apiKey: Resource.XAI_API_KEY.value,
- model: "grok-code",
- cost: {
- input: 0,
- output: 0,
- reasoning: 0,
- cacheRead: 0,
- cacheWrite: 0,
- },
- headerMappings: {
- "x-grok-conv-id": "x-opencode-session",
- "x-grok-req-id": "x-opencode-request",
- },
- },
- // deprecated
- "qwen/qwen3-coder": {
- id: "qwen/qwen3-coder" as const,
- auth: true,
- api: "https://inference.baseten.co",
- apiKey: Resource.BASETEN_API_KEY.value,
- model: "Qwen/Qwen3-Coder-480B-A35B-Instruct",
- cost: {
- input: 0.00000038,
- output: 0.00000153,
- reasoning: 0,
- cacheRead: 0,
- cacheWrite: 0,
- },
- headerMappings: {},
- },
-}
-
-const FREE_WORKSPACES = [
- "wrk_01K46JDFR0E75SG2Q8K172KF3Y", // frank
-]
-
-class AuthError extends Error {}
-class CreditsError extends Error {}
-class ModelError extends Error {}
-
-export async function POST(input: APIEvent) {
- try {
- const url = new URL(input.request.url)
- const body = await input.request.json()
- logMetric({
- is_tream: !!body.stream,
- session: input.request.headers.get("x-opencode-session"),
- request: input.request.headers.get("x-opencode-request"),
- })
- const MODEL = validateModel()
- const apiKey = await authenticate()
- const isFree = FREE_WORKSPACES.includes(apiKey?.workspaceID ?? "")
- await checkCredits()
-
- // Request to model provider
- const res = await fetch(new URL(url.pathname.replace(/^\/zen/, "") + url.search, MODEL.api), {
- method: "POST",
- headers: (() => {
- const headers = input.request.headers
- headers.delete("host")
- headers.delete("content-length")
- headers.set("authorization", `Bearer ${MODEL.apiKey}`)
- Object.entries(MODEL.headerMappings ?? {}).forEach(([k, v]) => {
- headers.set(k, headers.get(v)!)
- })
- return headers
- })(),
- body: JSON.stringify({
- ...body,
- model: MODEL.model,
- stream_options: {
- include_usage: true,
- },
- }),
- })
-
- // Scrub response headers
- const resHeaders = new Headers()
- const keepHeaders = ["content-type", "cache-control"]
- for (const [k, v] of res.headers.entries()) {
- if (keepHeaders.includes(k.toLowerCase())) {
- resHeaders.set(k, v)
- }
- }
-
- // Handle non-streaming response
- if (!body.stream) {
- const json = await res.json()
- const body = JSON.stringify(json)
- logMetric({ response_length: body.length })
- await trackUsage(json)
- return new Response(body, {
- status: res.status,
- statusText: res.statusText,
- headers: resHeaders,
- })
- }
-
- // Handle streaming response
- const stream = new ReadableStream({
- start(c) {
- const reader = res.body?.getReader()
- const decoder = new TextDecoder()
- let buffer = ""
- let responseLength = 0
- let startTimestamp = Date.now()
- let receivedFirstByte = false
-
- function pump(): Promise<void> {
- return (
- reader?.read().then(async ({ done, value }) => {
- if (done) {
- logMetric({ response_length: responseLength })
- c.close()
- return
- }
-
- if (!receivedFirstByte) {
- receivedFirstByte = true
- logMetric({ time_to_first_byte: Date.now() - startTimestamp })
- }
-
- buffer += decoder.decode(value, { stream: true })
- responseLength += value.length
-
- const parts = buffer.split("\n\n")
- buffer = parts.pop() ?? ""
-
- const usage = parts
- .map((part) => part.trim())
- .filter((part) => part.startsWith("data: "))
- .map((part) => {
- try {
- return JSON.parse(part.slice(6))
- } catch (e) {
- return {}
- }
- })
- .find((part) => part.usage)
- if (usage) await trackUsage(usage)
-
- c.enqueue(value)
-
- return pump()
- }) || Promise.resolve()
- )
- }
-
- return pump()
+import { handler } from "~/util/zen"
+
+export function POST(input: APIEvent) {
+ return handler(input, {
+ transformBody: (body: any) => ({
+ ...body,
+ stream_options: {
+ include_usage: true,
},
- })
-
- return new Response(stream, {
- status: res.status,
- statusText: res.statusText,
- headers: resHeaders,
- })
-
- function validateModel() {
- if (!(body.model in MODELS)) {
- throw new ModelError(`Model ${body.model} not supported`)
- }
- const model = MODELS[body.model as keyof typeof MODELS]
- logMetric({ model: model.id })
- return model
- }
+ }),
+ parseUsageChunk: (chunk: string) => {
+ if (!chunk.startsWith("data: ")) return
- async function authenticate() {
+ let json
try {
- const authHeader = input.request.headers.get("authorization")
- if (!authHeader || !authHeader.startsWith("Bearer ")) throw new AuthError("Missing API key.")
-
- const apiKey = authHeader.split(" ")[1]
- const key = await Database.use((tx) =>
- tx
- .select({
- id: KeyTable.id,
- workspaceID: KeyTable.workspaceID,
- })
- .from(KeyTable)
- .where(eq(KeyTable.key, apiKey))
- .then((rows) => rows[0]),
- )
-
- if (!key) throw new AuthError("Invalid API key.")
- logMetric({
- api_key: key.id,
- workspace: key.workspaceID,
- })
- return key
+ json = JSON.parse(chunk.slice(6))
} catch (e) {
- // ignore error if model does not require authentication
- if (!MODEL.auth) return
- throw e
+ return
}
- }
-
- async function checkCredits() {
- if (!apiKey || !MODEL.auth || isFree) return
-
- const billing = await Database.use((tx) =>
- tx
- .select({
- balance: BillingTable.balance,
- })
- .from(BillingTable)
- .where(eq(BillingTable.workspaceID, apiKey.workspaceID))
- .then((rows) => rows[0]),
- )
- if (billing.balance <= 0) throw new CreditsError("Insufficient balance")
- }
-
- async function trackUsage(chunk: any) {
- const usage = chunk.usage
- const inputTokens = usage.prompt_tokens ?? 0
- const outputTokens = usage.completion_tokens ?? 0
- const reasoningTokens = usage.completion_tokens_details?.reasoning_tokens ?? 0
- const cacheReadTokens = usage.prompt_tokens_details?.cached_tokens ?? 0
- //const cacheWriteTokens = providerMetadata?.["anthropic"]?.["cacheCreationInputTokens"] ?? 0
- const cacheWriteTokens = 0
-
- const inputCost = MODEL.cost.input * inputTokens * 100
- const outputCost = MODEL.cost.output * outputTokens * 100
- const reasoningCost = MODEL.cost.reasoning * reasoningTokens * 100
- const cacheReadCost = MODEL.cost.cacheRead * cacheReadTokens * 100
- const cacheWriteCost = MODEL.cost.cacheWrite * cacheWriteTokens * 100
- const totalCostInCent = inputCost + outputCost + reasoningCost + cacheReadCost + cacheWriteCost
-
- logMetric({
- "tokens.input": inputTokens,
- "tokens.output": outputTokens,
- "tokens.reasoning": reasoningTokens,
- "tokens.cache_read": cacheReadTokens,
- "tokens.cache_write": cacheWriteTokens,
- "cost.input": Math.round(inputCost),
- "cost.output": Math.round(outputCost),
- "cost.reasoning": Math.round(reasoningCost),
- "cost.cache_read": Math.round(cacheReadCost),
- "cost.cache_write": Math.round(cacheWriteCost),
- "cost.total": Math.round(totalCostInCent),
- })
-
- if (!apiKey) return
-
- const cost = isFree ? 0 : centsToMicroCents(totalCostInCent)
- await Database.transaction(async (tx) => {
- await tx.insert(UsageTable).values({
- workspaceID: apiKey.workspaceID,
- id: Identifier.create("usage"),
- model: MODEL.id,
- inputTokens,
- outputTokens,
- reasoningTokens,
- cacheReadTokens,
- cacheWriteTokens,
- cost,
- })
- await tx
- .update(BillingTable)
- .set({
- balance: sql`${BillingTable.balance} - ${cost}`,
- })
- .where(eq(BillingTable.workspaceID, apiKey.workspaceID))
- })
-
- await Database.use((tx) =>
- tx
- .update(KeyTable)
- .set({ timeUsed: sql`now()` })
- .where(eq(KeyTable.id, apiKey.id)),
- )
- }
- } catch (error: any) {
- logMetric({
- "error.type": error.constructor.name,
- "error.message": error.message,
- })
-
- if (error instanceof AuthError || error instanceof CreditsError || error instanceof ModelError)
- return new Response(JSON.stringify({ error: { message: error.message } }), { status: 401 })
-
- return new Response(JSON.stringify({ error: { message: error.message } }), { status: 500 })
- }
-
- function logMetric(values: Record<string, any>) {
- console.log(`_metric:${JSON.stringify(values)}`)
- }
+ return json.usage
+ },
+ buildUsage: (usage: any) => ({
+ inputTokens: usage.prompt_tokens ?? 0,
+ outputTokens: usage.completion_tokens ?? 0,
+ reasoningTokens: usage.completion_tokens_details?.reasoning_tokens ?? 0,
+ cacheReadTokens: usage.prompt_tokens_details?.cached_tokens ?? 0,
+ //cacheWriteTokens = usage.providerMetadata?.["anthropic"]?.["cacheCreationInputTokens"] ?? 0
+ cacheWriteTokens: 0,
+ }),
+ })
}
diff --git a/cloud/app/src/routes/zen/v1/responses.ts b/cloud/app/src/routes/zen/v1/responses.ts
new file mode 100644
index 000000000..844b4fef8
--- /dev/null
+++ b/cloud/app/src/routes/zen/v1/responses.ts
@@ -0,0 +1,34 @@
+import type { APIEvent } from "@solidjs/start/server"
+import { handler } from "~/util/zen"
+
+export function POST(input: APIEvent) {
+ return handler(input, {
+ parseUsageChunk: (chunk: string) => {
+ const [event, data] = chunk.split("\n")
+ if (event !== "event: response.completed") return
+ if (!data.startsWith("data: ")) return
+
+ let json
+ try {
+ json = JSON.parse(data.slice(6))
+ } catch (e) {
+ return
+ }
+
+ return json.response?.usage
+ },
+ buildUsage: (usage: any) => {
+ const inputTokens = usage.input_tokens ?? 0
+ const outputTokens = usage.output_tokens ?? 0
+ const reasoningTokens = usage.output_tokens_details?.reasoning_tokens ?? 0
+ const cacheReadTokens = usage.input_tokens_details?.cached_tokens ?? 0
+ return {
+ inputTokens: inputTokens - cacheReadTokens,
+ outputTokens: outputTokens - reasoningTokens,
+ reasoningTokens,
+ cacheReadTokens,
+ cacheWriteTokens: 0,
+ }
+ },
+ })
+}
diff --git a/cloud/app/src/util/zen.ts b/cloud/app/src/util/zen.ts
new file mode 100644
index 000000000..b2066e19b
--- /dev/null
+++ b/cloud/app/src/util/zen.ts
@@ -0,0 +1,367 @@
+import type { APIEvent } from "@solidjs/start/server"
+import { Database, eq, sql } from "@opencode/cloud-core/drizzle/index.js"
+import { KeyTable } from "@opencode/cloud-core/schema/key.sql.js"
+import { BillingTable, UsageTable } from "@opencode/cloud-core/schema/billing.sql.js"
+import { centsToMicroCents } from "@opencode/cloud-core/util/price.js"
+import { Identifier } from "@opencode/cloud-core/identifier.js"
+import { Resource } from "@opencode/cloud-resource"
+
+class AuthError extends Error {}
+class CreditsError extends Error {}
+class ModelError extends Error {}
+
+const MODELS = {
+ // "anthropic/claude-sonnet-4": {
+ // auth: true,
+ // api: "https://api.anthropic.com",
+ // apiKey: Resource.ANTHROPIC_API_KEY.value,
+ // model: "claude-sonnet-4-20250514",
+ // cost: {
+ // input: 0.0000015,
+ // output: 0.000006,
+ // reasoning: 0.0000015,
+ // cacheRead: 0.0000001,
+ // cacheWrite: 0.0000001,
+ // },
+ // headerMappings: {},
+ // },
+ "gpt-5": {
+ id: "gpt-5" as const,
+ auth: true,
+ api: "https://api.openai.com",
+ apiKey: Resource.OPENAI_API_KEY.value,
+ model: "gpt-5",
+ cost: {
+ input: 0.00000125,
+ output: 0.00001,
+ reasoning: 0.00001,
+ cacheRead: 0.000000125,
+ cacheWrite: 0,
+ },
+ headerMappings: {},
+ },
+ "qwen3-coder": {
+ id: "qwen3-coder" as const,
+ auth: true,
+ api: "https://inference.baseten.co",
+ apiKey: Resource.BASETEN_API_KEY.value,
+ model: "Qwen/Qwen3-Coder-480B-A35B-Instruct",
+ cost: {
+ input: 0.00000038,
+ output: 0.00000153,
+ reasoning: 0,
+ cacheRead: 0,
+ cacheWrite: 0,
+ },
+ headerMappings: {},
+ },
+ "kimi-k2": {
+ id: "kimi-k2" as const,
+ auth: true,
+ api: "https://inference.baseten.co",
+ apiKey: Resource.BASETEN_API_KEY.value,
+ model: "moonshotai/Kimi-K2-Instruct-0905",
+ cost: {
+ input: 0.0000006,
+ output: 0.0000025,
+ reasoning: 0,
+ cacheRead: 0,
+ cacheWrite: 0,
+ },
+ headerMappings: {},
+ },
+ "grok-code": {
+ id: "grok-code" as const,
+ auth: false,
+ api: "https://api.x.ai",
+ apiKey: Resource.XAI_API_KEY.value,
+ model: "grok-code",
+ cost: {
+ input: 0,
+ output: 0,
+ reasoning: 0,
+ cacheRead: 0,
+ cacheWrite: 0,
+ },
+ headerMappings: {
+ "x-grok-conv-id": "x-opencode-session",
+ "x-grok-req-id": "x-opencode-request",
+ },
+ },
+ // deprecated
+ "qwen/qwen3-coder": {
+ id: "qwen/qwen3-coder" as const,
+ auth: true,
+ api: "https://inference.baseten.co",
+ apiKey: Resource.BASETEN_API_KEY.value,
+ model: "Qwen/Qwen3-Coder-480B-A35B-Instruct",
+ cost: {
+ input: 0.00000038,
+ output: 0.00000153,
+ reasoning: 0,
+ cacheRead: 0,
+ cacheWrite: 0,
+ },
+ headerMappings: {},
+ },
+}
+
+const FREE_WORKSPACES = [
+ "wrk_01K46JDFR0E75SG2Q8K172KF3Y", // frank
+]
+
+const logger = {
+ metric: (values: Record<string, any>) => {
+ console.log(`_metric:${JSON.stringify(values)}`)
+ },
+ log: console.log,
+ debug: (message: string) => {
+ if (Resource.App.stage === "production") return
+ console.debug(message)
+ },
+}
+
+export async function handler(
+ input: APIEvent,
+ opts: {
+ transformBody?: (body: any) => any
+ parseUsageChunk: (chunk: string) => string | undefined
+ buildUsage: (body: any) => {
+ inputTokens: number
+ outputTokens: number
+ reasoningTokens: number
+ cacheReadTokens: number
+ cacheWriteTokens: number
+ }
+ },
+) {
+ try {
+ const url = new URL(input.request.url)
+ const body = await input.request.json()
+ logger.debug(JSON.stringify(body))
+ logger.metric({
+ is_tream: !!body.stream,
+ session: input.request.headers.get("x-opencode-session"),
+ request: input.request.headers.get("x-opencode-request"),
+ })
+ const MODEL = validateModel()
+ const apiKey = await authenticate()
+ const isFree = FREE_WORKSPACES.includes(apiKey?.workspaceID ?? "")
+ await checkCredits()
+
+ // Request to model provider
+ const res = await fetch(new URL(url.pathname.replace(/^\/zen/, "") + url.search, MODEL.api), {
+ method: "POST",
+ headers: (() => {
+ const headers = input.request.headers
+ headers.delete("host")
+ headers.delete("content-length")
+ headers.set("authorization", `Bearer ${MODEL.apiKey}`)
+ Object.entries(MODEL.headerMappings ?? {}).forEach(([k, v]) => {
+ headers.set(k, headers.get(v)!)
+ })
+ return headers
+ })(),
+ body: JSON.stringify({
+ ...(opts.transformBody?.(body) ?? body),
+ model: MODEL.model,
+ }),
+ })
+
+ // Scrub response headers
+ const resHeaders = new Headers()
+ const keepHeaders = ["content-type", "cache-control"]
+ for (const [k, v] of res.headers.entries()) {
+ if (keepHeaders.includes(k.toLowerCase())) {
+ resHeaders.set(k, v)
+ }
+ }
+
+ // Handle non-streaming response
+ if (!body.stream) {
+ const json = await res.json()
+ const body = JSON.stringify(json)
+ logger.metric({ response_length: body.length })
+ logger.debug(body)
+ await trackUsage(json.usage)
+ return new Response(body, {
+ status: res.status,
+ statusText: res.statusText,
+ headers: resHeaders,
+ })
+ }
+
+ // Handle streaming response
+ const stream = new ReadableStream({
+ start(c) {
+ const reader = res.body?.getReader()
+ const decoder = new TextDecoder()
+ let buffer = ""
+ let responseLength = 0
+ let startTimestamp = Date.now()
+ let receivedFirstByte = false
+
+ function pump(): Promise<void> {
+ return (
+ reader?.read().then(async ({ done, value }) => {
+ if (done) {
+ logger.metric({ response_length: responseLength })
+ c.close()
+ return
+ }
+
+ if (!receivedFirstByte) {
+ receivedFirstByte = true
+ logger.metric({ time_to_first_byte: Date.now() - startTimestamp })
+ }
+
+ buffer += decoder.decode(value, { stream: true })
+ responseLength += value.length
+
+ const parts = buffer.split("\n\n")
+ buffer = parts.pop() ?? ""
+
+ for (const part of parts) {
+ logger.debug(part)
+ const usage = opts.parseUsageChunk(part.trim())
+ if (usage) await trackUsage(usage)
+ }
+
+ c.enqueue(value)
+
+ return pump()
+ }) || Promise.resolve()
+ )
+ }
+
+ return pump()
+ },
+ })
+
+ return new Response(stream, {
+ status: res.status,
+ statusText: res.statusText,
+ headers: resHeaders,
+ })
+
+ function validateModel() {
+ if (!(body.model in MODELS)) {
+ throw new ModelError(`Model ${body.model} not supported`)
+ }
+ const model = MODELS[body.model as keyof typeof MODELS]
+ logger.metric({ model: model.id })
+ return model
+ }
+
+ async function authenticate() {
+ try {
+ const authHeader = input.request.headers.get("authorization")
+ if (!authHeader || !authHeader.startsWith("Bearer ")) throw new AuthError("Missing API key.")
+
+ const apiKey = authHeader.split(" ")[1]
+ const key = await Database.use((tx) =>
+ tx
+ .select({
+ id: KeyTable.id,
+ workspaceID: KeyTable.workspaceID,
+ })
+ .from(KeyTable)
+ .where(eq(KeyTable.key, apiKey))
+ .then((rows) => rows[0]),
+ )
+
+ if (!key) throw new AuthError("Invalid API key.")
+ logger.metric({
+ api_key: key.id,
+ workspace: key.workspaceID,
+ })
+ return key
+ } catch (e) {
+ // ignore error if model does not require authentication
+ if (!MODEL.auth) return
+ throw e
+ }
+ }
+
+ async function checkCredits() {
+ if (!apiKey || !MODEL.auth || isFree) return
+
+ const billing = await Database.use((tx) =>
+ tx
+ .select({
+ balance: BillingTable.balance,
+ })
+ .from(BillingTable)
+ .where(eq(BillingTable.workspaceID, apiKey.workspaceID))
+ .then((rows) => rows[0]),
+ )
+
+ if (billing.balance <= 0) throw new CreditsError("Insufficient balance")
+ }
+
+ async function trackUsage(usage: any) {
+ const { inputTokens, outputTokens, reasoningTokens, cacheReadTokens, cacheWriteTokens } = opts.buildUsage(usage)
+
+ const inputCost = MODEL.cost.input * inputTokens * 100
+ const outputCost = MODEL.cost.output * outputTokens * 100
+ const reasoningCost = MODEL.cost.reasoning * reasoningTokens * 100
+ const cacheReadCost = MODEL.cost.cacheRead * cacheReadTokens * 100
+ const cacheWriteCost = MODEL.cost.cacheWrite * cacheWriteTokens * 100
+ const totalCostInCent = inputCost + outputCost + reasoningCost + cacheReadCost + cacheWriteCost
+
+ logger.metric({
+ "tokens.input": inputTokens,
+ "tokens.output": outputTokens,
+ "tokens.reasoning": reasoningTokens,
+ "tokens.cache_read": cacheReadTokens,
+ "tokens.cache_write": cacheWriteTokens,
+ "cost.input": Math.round(inputCost),
+ "cost.output": Math.round(outputCost),
+ "cost.reasoning": Math.round(reasoningCost),
+ "cost.cache_read": Math.round(cacheReadCost),
+ "cost.cache_write": Math.round(cacheWriteCost),
+ "cost.total": Math.round(totalCostInCent),
+ })
+
+ if (!apiKey) return
+
+ const cost = isFree ? 0 : centsToMicroCents(totalCostInCent)
+ await Database.transaction(async (tx) => {
+ await tx.insert(UsageTable).values({
+ workspaceID: apiKey.workspaceID,
+ id: Identifier.create("usage"),
+ model: MODEL.id,
+ inputTokens,
+ outputTokens,
+ reasoningTokens,
+ cacheReadTokens,
+ cacheWriteTokens,
+ cost,
+ })
+ await tx
+ .update(BillingTable)
+ .set({
+ balance: sql`${BillingTable.balance} - ${cost}`,
+ })
+ .where(eq(BillingTable.workspaceID, apiKey.workspaceID))
+ })
+
+ await Database.use((tx) =>
+ tx
+ .update(KeyTable)
+ .set({ timeUsed: sql`now()` })
+ .where(eq(KeyTable.id, apiKey.id)),
+ )
+ }
+ } catch (error: any) {
+ logger.metric({
+ "error.type": error.constructor.name,
+ "error.message": error.message,
+ })
+
+ if (error instanceof AuthError || error instanceof CreditsError || error instanceof ModelError)
+ return new Response(JSON.stringify({ error: { message: error.message } }), { status: 401 })
+
+ return new Response(JSON.stringify({ error: { message: error.message } }), { status: 500 })
+ }
+}
diff --git a/cloud/function/sst-env.d.ts b/cloud/function/sst-env.d.ts
index efb047ff5..fd96591b8 100644
--- a/cloud/function/sst-env.d.ts
+++ b/cloud/function/sst-env.d.ts
@@ -54,6 +54,10 @@ declare module "sst" {
"type": "sst.sst.Secret"
"value": string
}
+ "OPENAI_API_KEY": {
+ "type": "sst.sst.Secret"
+ "value": string
+ }
"STRIPE_SECRET_KEY": {
"type": "sst.sst.Secret"
"value": string
diff --git a/cloud/resource/resource.cloudflare.ts b/cloud/resource/resource.cloudflare.ts
index ad3fbe2a8..a56b1e417 100644
--- a/cloud/resource/resource.cloudflare.ts
+++ b/cloud/resource/resource.cloudflare.ts
@@ -8,6 +8,9 @@ export const Resource = new Proxy(
// @ts-expect-error
const value = env[prop]
return typeof value === "string" ? JSON.parse(value) : value
+ } else if (prop === "App") {
+ // @ts-expect-error
+ return JSON.parse(env.SST_RESOURCE_App)
}
throw new Error(`"${prop}" is not linked in your sst.config.ts (cloudflare)`)
},
diff --git a/cloud/resource/sst-env.d.ts b/cloud/resource/sst-env.d.ts
index efb047ff5..fd96591b8 100644
--- a/cloud/resource/sst-env.d.ts
+++ b/cloud/resource/sst-env.d.ts
@@ -54,6 +54,10 @@ declare module "sst" {
"type": "sst.sst.Secret"
"value": string
}
+ "OPENAI_API_KEY": {
+ "type": "sst.sst.Secret"
+ "value": string
+ }
"STRIPE_SECRET_KEY": {
"type": "sst.sst.Secret"
"value": string
diff --git a/infra/cloud.ts b/infra/cloud.ts
index 9cb5a83f1..622068e08 100644
--- a/infra/cloud.ts
+++ b/infra/cloud.ts
@@ -100,6 +100,7 @@ export const stripeWebhook = new WebhookEndpoint("StripeWebhookEndpoint", {
})
const ANTHROPIC_API_KEY = new sst.Secret("ANTHROPIC_API_KEY")
+const OPENAI_API_KEY = new sst.Secret("OPENAI_API_KEY")
const XAI_API_KEY = new sst.Secret("XAI_API_KEY")
const BASETEN_API_KEY = new sst.Secret("BASETEN_API_KEY")
const STRIPE_SECRET_KEY = new sst.Secret("STRIPE_SECRET_KEY")
diff --git a/packages/function/sst-env.d.ts b/packages/function/sst-env.d.ts
index efb047ff5..fd96591b8 100644
--- a/packages/function/sst-env.d.ts
+++ b/packages/function/sst-env.d.ts
@@ -54,6 +54,10 @@ declare module "sst" {
"type": "sst.sst.Secret"
"value": string
}
+ "OPENAI_API_KEY": {
+ "type": "sst.sst.Secret"
+ "value": string
+ }
"STRIPE_SECRET_KEY": {
"type": "sst.sst.Secret"
"value": string
diff --git a/packages/opencode/src/provider/models.ts b/packages/opencode/src/provider/models.ts
index 8cd03088f..e41221c2a 100644
--- a/packages/opencode/src/provider/models.ts
+++ b/packages/opencode/src/provider/models.ts
@@ -30,6 +30,7 @@ export namespace ModelsDev {
}),
experimental: z.boolean().optional(),
options: z.record(z.any()),
+ provider: z.object({ npm: z.string() }).optional(),
})
.openapi({
ref: "Model",
diff --git a/packages/opencode/src/provider/provider.ts b/packages/opencode/src/provider/provider.ts
index 592d0c023..dbed1f74d 100644
--- a/packages/opencode/src/provider/provider.ts
+++ b/packages/opencode/src/provider/provider.ts
@@ -235,6 +235,7 @@ export namespace Provider {
context: 0,
output: 0,
},
+ provider: model.provider ?? existing?.provider,
}
parsed.models[modelID] = parsedModel
}
@@ -319,7 +320,7 @@ export namespace Provider {
return state().then((state) => state.providers)
}
- async function getSDK(provider: ModelsDev.Provider) {
+ async function getSDK(provider: ModelsDev.Provider, model: ModelsDev.Model) {
return (async () => {
using _ = log.time("getSDK", {
providerID: provider.id,
@@ -327,7 +328,7 @@ export namespace Provider {
const s = await state()
const existing = s.sdk.get(provider.id)
if (existing) return existing
- const pkg = provider.npm ?? provider.id
+ const pkg = model.provider?.npm ?? provider.npm ?? provider.id
const mod = await import(await BunProc.install(pkg, "latest"))
const fn = mod[Object.keys(mod).find((key) => key.startsWith("create"))!]
let options = { ...s.providers[provider.id]?.options }
@@ -366,7 +367,7 @@ export namespace Provider {
if (!provider) throw new ModelNotFoundError({ providerID, modelID })
const info = provider.info.models[modelID]
if (!info) throw new ModelNotFoundError({ providerID, modelID })
- const sdk = await getSDK(provider.info)
+ const sdk = await getSDK(provider.info, info)
try {
const language = provider.getModel ? await provider.getModel(sdk, modelID) : sdk.languageModel(modelID)
diff --git a/sst-env.d.ts b/sst-env.d.ts
index c1999089e..6392ab739 100644
--- a/sst-env.d.ts
+++ b/sst-env.d.ts
@@ -71,6 +71,10 @@ declare module "sst" {
"LogProcessor": {
"type": "sst.cloudflare.Worker"
}
+ "OPENAI_API_KEY": {
+ "type": "sst.sst.Secret"
+ "value": string
+ }
"STRIPE_SECRET_KEY": {
"type": "sst.sst.Secret"
"value": string
generated by cgit v1.2.3 (git 2.39.1) at 2026-06-29 12:20:45 +0000