chore: switch model from Claude Opus 4.7 to 4.6 (EU available)
This commit is contained in:
parent
25847af026
commit
32c9c78b42
|
|
@ -1,6 +1,6 @@
|
||||||
# GEMINI.md — OSVauco / OPAX
|
# GEMINI.md — OSVauco / OPAX
|
||||||
|
|
||||||
> Claude Opus 4.7 · Vertex AI EU endpoint · Nemotron session protocol · HITL-safe
|
> Claude Opus 4.6 · Vertex AI EU endpoint · Nemotron session protocol · HITL-safe
|
||||||
|
|
||||||
---
|
---
|
||||||
|
|
||||||
|
|
@ -16,7 +16,7 @@ GCP_REGION: europe-west1
|
||||||
GCP_VERTEX_LOC: eu
|
GCP_VERTEX_LOC: eu
|
||||||
CLOUD_RUN_SVC: osvauco-agent
|
CLOUD_RUN_SVC: osvauco-agent
|
||||||
ARTIFACT_REPO: gcr.io/propane-will-491900-m5
|
ARTIFACT_REPO: gcr.io/propane-will-491900-m5
|
||||||
MODEL: claude-opus-4-7@20260301
|
MODEL: claude-opus-4-6@20260801
|
||||||
MODEL_PROVIDER: Vertex AI (Anthropic Claude via Google Cloud)
|
MODEL_PROVIDER: Vertex AI (Anthropic Claude via Google Cloud)
|
||||||
MODEL_COST: ~2 NOK per 1M tokens (input)
|
MODEL_COST: ~2 NOK per 1M tokens (input)
|
||||||
BUDGET_CAP: 2500 NOK intern grense — spør Chris om +1000 NOK ved behov
|
BUDGET_CAP: 2500 NOK intern grense — spør Chris om +1000 NOK ved behov
|
||||||
|
|
@ -28,9 +28,9 @@ OPERATOR_ROLE: Eier, arkitekt og primær operatør av OSVauco / OS-Vauco
|
||||||
|
|
||||||
---
|
---
|
||||||
|
|
||||||
## Modell: Claude Opus 4.7 via Vertex AI
|
## Modell: Claude Opus 4.6 via Vertex AI
|
||||||
|
|
||||||
Denne agenten kjøres med **Claude Opus 4.7** tilgjengelig via Vertex AI EU multi-region endpoint.
|
Denne agenten kjøres med **Claude Opus 4.6** tilgjengelig via Vertex AI EU multi-region endpoint.
|
||||||
|
|
||||||
- Data forblir innenfor EU-grensen (GDPR-compliant)
|
- Data forblir innenfor EU-grensen (GDPR-compliant)
|
||||||
- Ingen separat Anthropic API-nøkkel — billing via eksisterende GCP-konto
|
- Ingen separat Anthropic API-nøkkel — billing via eksisterende GCP-konto
|
||||||
|
|
@ -47,7 +47,7 @@ client = anthropic.AnthropicVertex(
|
||||||
)
|
)
|
||||||
|
|
||||||
response = client.messages.create(
|
response = client.messages.create(
|
||||||
model="claude-opus-4-7@20260301",
|
model="claude-opus-4-6@20260801",
|
||||||
max_tokens=8192,
|
max_tokens=8192,
|
||||||
messages=[{"role": "user", "content": "..."}]
|
messages=[{"role": "user", "content": "..."}]
|
||||||
)
|
)
|
||||||
|
|
@ -321,9 +321,9 @@ gcloud iap web get-iam-policy \
|
||||||
### 🔴 Claude / Vertex AI-feil
|
### 🔴 Claude / Vertex AI-feil
|
||||||
|
|
||||||
```bash
|
```bash
|
||||||
# Test Claude Opus 4.7 via Vertex EU endpoint
|
# Test Claude Opus 4.6 via Vertex EU endpoint
|
||||||
curl -X POST \
|
curl -X POST \
|
||||||
"https://eu-aiplatform.googleapis.com/v1/projects/propane-will-491900-m5/locations/eu/publishers/anthropic/models/claude-opus-4-7@20260301:rawPredict" \
|
"https://eu-aiplatform.googleapis.com/v1/projects/propane-will-491900-m5/locations/eu/publishers/anthropic/models/claude-opus-4-6@20260801:rawPredict" \
|
||||||
-H "Authorization: Bearer $(gcloud auth print-access-token)" \
|
-H "Authorization: Bearer $(gcloud auth print-access-token)" \
|
||||||
-H "Content-Type: application/json" \
|
-H "Content-Type: application/json" \
|
||||||
-d '{
|
-d '{
|
||||||
|
|
|
||||||
Loading…
Reference in New Issue
Block a user