LLM API Daily · 2026-05-31
Scanned: 47 entries · 0 breaking · 0 deprecations
🚨 Breaking
Nothing to report today.
🗑️ Deprecations
Nothing to report today.
💰 Pricing
Nothing to report today.
🆕 New
Anthropic — Managed Agents now on AWS
Claude Managed Agents webhooks, multiagent orchestration, and self-hosted sandboxes are now available on Claude Platform on AWS. New IAM actions and the AnthropicSelfHostedEnvironmentAccess managed policy have been documented.
→ Anthropic API release notes, May 29 2026
llama.cpp b9437 — -fa auto now supported in llama-bench; default -ngl changed to -1, aligning it with other llama.cpp tooling.
→ b9437
llama.cpp b9436 — OpenCL backend gains bf16 support (internally converted to f16).
→ b9436
llama.cpp b9414 — DeepSeekOCR 2 added with multi-tile dynamic resolution support.
→ b9414
llama.cpp b9415 — New skip_download flag: if the target file already exists, the download is skipped. Useful for idempotent model-management scripts.
→ b9415
llama.cpp b9430 — LoongArch: native LSX fp16 load/store intrinsics and LSX dot-product implementations for q8_0 and q6_K quantizations.
→ b9430
🌐 AI Industry
Google I/O 2026 — Google published 9 demos of Gemini Omni and Gemini 3.5 in action, covering live multimodal capabilities.
→ Gemini Omni & 3.5 demos
💡 Action Today
If you run workloads on Claude Platform on AWS: review the new IAM actions for Managed Agents and explicitly grant the AnthropicSelfHostedEnvironmentAccess managed policy before your next deploy — webhooks and multiagent orchestration endpoints are unreachable without it.