Fresh stories
Zyphra prices GLM-5.3-Flash FP4 inference 25% below FP8
Zyphra prices FP4 inference on MI300X GPUs 25% below its FP8 offering. AgentX tests report 9.5× faster p90 generation than stock SGLang, with about 95% of input tokens served from cache.
QuixiAI releases MIT-licensed QuixiGate Rust gateway
QuixiGate ships as a cross-platform Rust binary for routing requests to cloud and local models. It supports OAuth subscriptions, telemetry integrations and configurable fallback when usage limits are reached.
Orosz identifies database monitoring gaps in Revolut’s five-hour card outage
Gergely Orosz’s analysis identifies missing database-node monitoring and fragile quorum placement in Revolut’s five-hour card outage. He also questions recovery tools that depended on the failed GCP zone.
Zyphra prices GLM-5.3-Flash FP4 inference 25% below FP8
Zyphra prices FP4 inference on MI300X GPUs 25% below its FP8 offering. AgentX tests report 9.5× faster p90 generation than stock SGLang, with about 95% of input tokens served from cache.
CodeClimb checkpoints coding experiments through API-credit pauses of up to three days
CodeClimb’s builder describes Inngest checkpoints that preserve coding experiments through pauses of up to three days. Separate coding and evaluation sandboxes use scores and reviewer feedback to accept or revert changes.
Top storiesthis week
Claude Code adds project chat controls for parallel-thread status
Claude Code’s Projects coordinator now answers questions and reports parallel-thread status in project chat. Updates also add reminders, read-only sharing, /btw and /fork support, and easier artifact navigation.
Inkling 2 Small uses four-bit training, Thinking Machines reports 8% and 20% savings
Thinking Machines says Inkling 2 Small uses four-bit forward passes throughout pretraining and reinforcement learning. It reports training-time savings of 8% and 20%, respectively, without a separate checkpoint quantization step.
llama.cpp ships its first stable major release
llama.cpp and ggml have reached their first major release. The team commits to broad platform support, while Cohere separately announced runtime support for its North Micro Vision architecture.
Report: Microsoft began mitigating its email outage after 20 hours
Gergely Orosz reports that Microsoft began mitigating its email outage after 20 hours. His reporting describes registry patches, rack reboots and traffic routing; the root cause remains unconfirmed pending a postmortem.
OpenAI launches ChatGPT Wallet for approved agent purchases
ChatGPT Wallet lets Dots complete approved purchases without exposing card details to agents. The rollout starts with US Pro subscribers through partnerships with American Express, Mastercard and Visa.