# Highlights - July 2026

Naming policy update: going forward, model identifiers will reference only the major version number under semantic versioning, for example GLM 5 rather than GLM 5.1 or GLM 5.2. Minor and patch releases will no longer be announced by email; they will still be recorded here in the changelog. Major version promotions, and any hard removal of a previously reachable model ID, will continue to trigger a customer notification regardless of whether the underlying change was major or minor.

Promoted inference-glm5 (GLM 5.2) from Sandbox Models to Active Models; the model is now generally available on the Phoeniqs Model Service. Hugging Face reference: zai-org/GLM-5.2. Pricing is unchanged from the previous minor release; context window is 131'072 tokens.

Through July 31, 2026, requests to inference-glm-51-754b will be automatically redirected to inference-glm5. No customer-side change is required during this window.

Starting August 1, 2026, inference-glm-51-754b will be fully decommissioned and the redirect removed; see Decommissioned Models. Any application or agent still calling inference-glm-51-754b directly must be updated to inference-glm5 before then, or requests will start failing.

New Model Service Base LLM API plan (CHF 50/month, 50 credits) replaces Exploratory. See Subscription Plans.

New Additional LLM Credits top-up plan: add credits self-service in the Phoeniqs Console, no service ticket required. See How to add LLM Credits.

Updated MaaS subscription docs, FAQs, and guides; added model status dashboard link.

Added an IaaS Quickstart Guide with step-by-step portal purchase flow for Resource Groups, Capacity Pools, resources, and choosing between a Namespace or Hosted Cluster.

Expanded Key Management FAQ with new Cloud HSM questions covering HA scaling, GREP11/mTLS authentication, SLA, key ceremonies, and sandbox trials.

Added Connect MCP Tools guides for linking Phoeniqs Chat to enterprise systems via MCP, including Atlassian Rovo and Microsoft 365.