Announcements
Latest feature releases, service launches, and updates across major cloud providers from the last 30 days. Curious where these providers actually run? Explore Datacenters for their global footprint, or visit Compliance to verify their certifications. Want to check data freshness? See the platform Status. Last updated: Aug 20, 2026.
Tracked announcements by provider
Counts reflect the announcements logged in the last 30 days. See the release notes for the complete list.Select one or multiple options above to filter cloud announcements and updates.
Amazon Bedrock adds support for Claude 3.5 Sonnet & Haiku
Anthropic’s latest flagship reasoning and coding models are now generally available across multiple AWS regions with cross-region inference.
Azure AI Foundry expands DeepSeek-R1 & V3 Availability
Microsoft announces broader regional availability and increased serverless throughput limits for DeepSeek models deployed on Azure AI.
Cloudflare Workers AI introduces Serverless Fine-Tuned Model Inference
Run custom LoRA fine-tuned adapters on serverless edge GPUs globally with sub-50ms cold starts and automatic region routing.
Cloud SQL for PostgreSQL 17 is generally available
Google Cloud SQL now officially supports PostgreSQL 17, bringing significant performance improvements, query tuning, and logical replication enhancements.
OpenAI o3-mini reasoning model released with structured JSON output
High-speed reasoning model tailored for science, coding, and multi-step agentic planning with 60% lower token cost than full-tier models.
OCI Generative AI Agents support Llama 3.3 and Multimodal RAG
Oracle Cloud Infrastructure introduces Llama 3.3 foundational models to its Generative AI Agents service with integrated in-memory vector search.
Pinecone Serverless adds high-cardinality metadata indexing
Filter vector queries across billions of multi-tenant partitions with sub-5ms latency and zero dedicated cluster idle cost.
Cloudflare Hyperdrive adds caching acceleration for Redis & Valkey
Turn centralized databases and in-memory caches into globally distributed assets by pooling connections at the edge nearest to end-users.
New Premium CPU Droplets with Gen4 NVMe Storage
DigitalOcean launches next-generation Premium CPU Droplets featuring high-IOPS local NVMe storage and 10 Gbps enhanced network throughput.
Anthropic Claude API adds 5-minute Prompt Caching with 90% cost reduction
Cache repetitive context windows (codebases, documentation, long system instructions) to accelerate inference latency and dramatically cut input pricing.
Qwen-Max 2.5 flagship model upgrades in Model Studio
Alibaba Cloud releases significant updates to the Qwen-Max language model, improving mathematical reasoning and multi-turn coding execution.
Mistral Large updated with 128k context window and enhanced function calling
Mistral’s flagship model improves multilingual reasoning, Python coding benchmarks, and native JSON schema output conformance.
AWS Lambda introduces Graviton4 processor support
AWS Lambda functions can now be powered by AWS Graviton4 processors, delivering up to 30% better compute price-performance over previous generations.
Qdrant Cloud introduces dynamic quantization and distributed sharding
Scale vector indexes to billions of embeddings with 4x memory compression and distributed query replication across cloud providers.
Azure Functions support for Node.js 22 & Python 3.12
Node.js 22 LTS and Python 3.12 runtimes are now generally available on Azure Functions across consumption, premium, and dedicated hosting plans.
Magic Transit integrates unified Zero Trust Network Access (ZTNA)
DDoS protection, Next-Gen WAF, and private interconnect routing can now be provisioned with declarative Terraform configuration across all cloud boundaries.
Cohere Embed v3.5 with multi-vector hybrid search optimization
Enterprise embedding model optimized for retrieval-augmented generation (RAG), mitigating noise and delivering state-of-the-art MTEB accuracy.
Weaviate Cloud adds native multi-tenancy and offloading to S3/GCS
Automatically offload inactive vector indexes to object storage to reduce hot memory hosting costs by up to 80% for enterprise multi-tenant apps.
Aiven launches fully managed Valkey in-memory cache & database
Deploy open-source Redis alternative Valkey across AWS, Google Cloud, and Azure with automated backups, VPC peering, and 99.99% SLA.
Sources
The links below go to each provider's official release notes. We track a curated, comparable subset of updates above. This is not a complete list; verify directly with the provider.