Anthropic Hard Cutover and Databricks Summit Both Start Tomorrow

<p>Anthropic's <strong>June 15 double deadline</strong> hits tomorrow: pinned Claude 4 model IDs hard-fail at the API and Agent SDK billing splits into separate per-user credit pools — no grace period, no rollover. The <strong>Databricks Data+AI Summit</strong> (June 15–18) opens the same day with Lakebase serverless Postgres now GA on AWS and MLflow 3 ag...

Lead signals

Two platform deadlines arrive tomorrow. One summit opens.

  • Anthropic's June 15 hard cutover: pinned Claude 4 model IDs fail at the API with no grace period, and Agent SDK billing splits into a separate monthly credit pool. (Enterprise DNA)
  • Databricks Data+AI Summit opens at Moscone Center tomorrow (June 15–18); Lakebase serverless Postgres is now GA on AWS and MLflow 3 agent evaluation takes center stage across 800+ sessions. (Databricks)
  • Gemini 3.5 Pro — 2M-token context, Deep Think reasoning — remains in limited Vertex preview with no confirmed GA date, despite a June target announced at Google I/O. (TechTimes)
  • Open-weight code model competition sharpened June 12 with Qwen3 Coder Next and Kimi K2.7 Code both shipping, adding capable open alternatives for code-heavy agent workflows. (LLM Stats)
agent-sdk model-deprecation databricks-summit open-weights
Key signals

Key Signals

1 Anthropic's Double Cutover: Model IDs and Agent SDK Billing
June 15, 2026 — tomorrow
claude-sonnet-4-20250514 and claude-opus-4-20250514 hit end-of-life on the Anthropic API with no extension — requests return errors from June 15 onward. Simultaneously, Agent SDK and claude -p usage moves out of standard subscription limits into a separate per-user credit pool: $20/month on Pro, $100 on Max 5x, $200 on Max 20x. Credits do not roll over, and automation stops cold if the pool runs dry unless overflow billing is enabled. Both changes take effect on the same calendar day, making this the most operationally loaded cutover Anthropic has scheduled to date. (Enterprise DNA) (The New Stack)
2 Databricks Data+AI Summit Opens Tomorrow
June 14, 2026
More than 30,000 data and AI professionals converge on Moscone Center in San Francisco for the June 15–18 event. Databricks Lakebase — a fully managed, serverless Postgres with separated compute and storage — reached GA on AWS ahead of the summit, with Azure and Google Cloud expansions to follow. MLflow 3's mlflow.genai namespace, covering multi-turn agent tracing, synthetic evaluation, human feedback loops, and LLM judge scoring, is a summit focal point for teams building production agent quality controls. Keynotes include Microsoft CEO Satya Nadella and OpenAI co-founder Greg Brockman. (Databricks) (DBTA) (ChatForest)
3 Gemini 3.5 Pro June GA Remains Unconfirmed
as of June 14, 2026
Google's next frontier tier — 2M-token context, Deep Think reasoning, flagship multimodal — has been in limited Vertex preview since its Google I/O announcement in May. No precise GA date is set. Teams blocked on the context window or reasoning depth should not assume a specific launch day; the late-June window is still plausible but unconfirmed. (TechTimes)
Operator actions

Why It Matters / What To Watch

Builders with pinned Claude 4 model IDs have hours left
  • Migrating to claude-sonnet-4-6 or claude-opus-4-8 is a one-line string change; any team that hasn't done it faces broken production calls starting tomorrow morning. (Anthropic Deprecation Docs)
  • Agent SDK operators should confirm their monthly credit bucket size, audit which claude -p automations and GitHub Actions runs count against it, and decide now whether to enable overflow billing. (The New Stack)
Databricks summit is a live product announcement window this week
  • Lakebase GA makes Databricks a credible Postgres-native persistence layer for AI apps and RAG pipelines — watch for agent-framework integrations announced at the summit over the next four days. (DBTA)
  • MLflow 3 agent evaluation sessions are the practical operations story of the week: synthetic eval datasets, LLM-judge scoring, and online trace ingestion are the patterns production teams are standardizing on for agent quality control. (Databricks MLflow Docs)
Sources

Quick Links