feat(ai): switch from DeepSeek to local Ollama (qwen3:14b)

- Point AI client to localhost:11434/v1 (Ollama)
- Remove DeepSeek API key dependency
- Update model name to qwen3:14b across all routes
- Update test mocks to use createLocalLLMClient
- Removes need for DEEPSEEK_API_KEY setting

Co-Authored-By: Claude Haiku 4.5 <noreply@anthropic.com>
This commit is contained in:
2026-10-01 11:34:05 +08:00
co-authored by Claude Haiku 4.5
parent 69e510af74
commit 4f75dd5887
8 changed files with 24 additions and 40 deletions
+4 -6
View File
@@ -6,8 +6,7 @@ import { withUser, asAdmin } from '@/lib/db/with-user'
import { writeAuditLog } from '@/lib/db/audit'
import { incidents, sites, zones, auditLog } from '@/lib/db/schema'
import { eq, and, gte, desc } from 'drizzle-orm'
import { createDeepSeekClient } from '@/lib/claude/client'
import { getApiKey } from '@/lib/settings'
import { createLocalLLMClient } from '@/lib/claude/client'
type ZoneAggregate = {
zone: string
@@ -103,13 +102,12 @@ export async function POST() {
return NextResponse.json({ error: 'Rate limit: wait 60 seconds between AI requests' }, { status: 429 })
}
const deepseekKey = await getApiKey('DEEPSEEK_API_KEY')
const client = createDeepSeekClient(deepseekKey)
const client = createLocalLLMClient()
let res: Awaited<ReturnType<typeof client.chat.completions.create>>
try {
res = await client.chat.completions.create({
model: 'deepseek-v4-pro',
model: 'qwen3:14b',
max_tokens: 2048,
messages: [
{
@@ -168,7 +166,7 @@ Flag only zones with rising or elevated risk (at most 5 flags; do not flag healt
await withUser(session.sub, async tx =>
writeAuditLog(tx, 'incidents', session.sub, 'ai_risk_flags',
{ flags, summary: input.summary, model: 'deepseek-v4-pro' })
{ flags, summary: input.summary, model: 'qwen3:14b' })
)
return NextResponse.json({ flags, summary: input.summary })