`Claude Platform` on `AWS` Reaches General Availability

// related

• #0001

  #0001Models & APIOpenAI Python SDK4 days ago

  `openai-python` `v2.37.0` adds `service_tier` to compact responses

  40radar

  More control landed in the Python SDK's compact responses path, plus a small auth cleanup. Useful if you already depend on responses; otherwise this is a low-priority bump.

  • —service_tier now reaches the compact responses method, so latency/cost routing can stay consistent without dropping to a lower-level call.
  • —Pydantic iterators get eager validation support in internal types, which should catch bad streamed data earlier in typed pipelines.
  • —Workload identity auth no longer requires an unnecessary client_id; fewer auth edge cases for cloud deployments using federated identity.

  Source: github.com/openai/openai-python/releases/tag/v2.37.0Read original →
• #0002

  #0002Models & APIGeekNews6 days ago

  Claude shifts programmatic usage to monthly credits

  70radar

  Programmatic usage moves into a separate monthly credit pool on June 15. Easier to track Agent SDK automation, but the real per-task cost may change.

  • —Coverage includes Claude Agent SDK, claude -p, Claude Code GitHub Actions, and third-party apps built on the SDK. Billing changes across the whole automation stack at once.
  • —The date is locked to June 15, so any product with heavy CI, batch jobs, or agent runs should check current burn patterns before the switch.
  • —Separating chat-style plan usage from programmatic calls makes budgeting cleaner. It also exposes whether the current paid plan still makes economic sense.
  • —Third-party SDK wrappers are included too, so this does not get bypassed by using another tool layer. Hidden automation spend becomes easier to spot.

  Source: news.hada.io/topic?id=29494Read original →
• #0003

  #0003Models & APIAnthropic Python SDK6 days ago

  `anthropic-sdk-python` `v0.102.0` adds managed agent search block and cache diagnostics beta

  50radar

  anthropic-sdk-pythonPython SDK — typed access to Anthropic API features

  The Python SDK now exposes a managed-agents search result block and a beta for cache diagnostics. Useful if you ship against Anthropic APIs directly; otherwise this is a low-priority dependency bump.

  • —Adds BetaManagedAgentsSearchResultBlock types, so managed-agent search responses can be handled without custom typing glue.
  • —Introduces cache diagnostics beta support, which helps inspect cache behavior when tuning prompt reuse and token spend.
  • —pydantic iterator validation is now eager, reducing the chance of delayed type errors surfacing deep in runtime paths.
  • —No pricing, model, or API-surface reset here; this is a small SDK upgrade rather than a platform-level change.

  Source: github.com/anthropics/anthropic-sdk-python/releases/tag/Read original →