Skip to content
MustHave.ai
MustHave Hoodie Seal
THE MUSTHAVE NEWSROOM

AI news

What’s new. What matters. What do you think?

Top stories

By member votes

llm-d 1.0: Test Queue Fairness and Rolling Updates Under Load

llm-d 1.0 adds production controls for distributed inference. Test interactive deadlines, background throughput and session continuity during upgrades.

DeepInfer: What You Still Need for a Complete Inference Service

DeepSeek’s DeepInfer release supports offline inference on specified NVIDIA and Ascend hardware. Review the missing service components before planning deployment.

Cloudflare createAI: Check Routing and Credentials During Migration

Cloudflare createAI replaces two deprecated provider packages. Check model routes, API credentials, fallback behavior and the separate AI Search migration.

Cohere Rerank 5 Reaches OpenRouter: Compare Ranking and Context Focus

Cohere’s Embed 5 and Rerank 5 are now on OpenRouter. Keep the provider route, passage selection, and final-answer quality separate in an evaluation.

Gemini in Sheets Reaches Full BigQuery Data: Review the Query and Refresh

Gemini can analyze full connected BigQuery datasets from Sheets. Check eligible plans, query permissions, generated objects and refreshed results during rollout.

TorchNeuron Public Beta: Verify the Result After Moving a Model

TorchNeuron’s public beta brings a native PyTorch backend to Trainium. Check numerical agreement, CPU fallback, compilation and complete workload performance.

GitHub Copilot Model Policies: Check the Access Behind an API Update

GitHub’s model-policy APIs are generally available for enterprises, teams, apps and organizations. Verify delegation and effective user access after a policy change.

Cloudflare MCP Adds Access on Demand: Review Each New Grant

Cloudflare’s API MCP server supports step-up authorization. Start with minimal scopes, then review the extra access requested for a particular tool operation.

Cloudflare Clairic: A Valid Field Still Needs the Right Answer

Clairic is a Workers AI extraction-model preview with typed answers and confidence values. Test missing evidence, calculations and dates before automating decisions.

MiniMax M3.1 Flash: Price the Finished Task Before Choosing a Model

MiniMax M3.1 Flash is available through its API, M Plan and MiniMax Code. Check published token rates, completed-task costs and the status of model weights.

Cloudflare Cloud Workload Security: Check the API Proxy Boundary

Cloudflare’s Cloud Workload Security beta checks routed cloud API requests. Review enforcement, unmatched-request defaults and error handling before relying on it.

ChatGPT Slides: Review the Shared Deck and the Exported File

ChatGPT Slides adds collaborative decks in Space. Check access, desktop editing, sharing roles and the exported PPTX, PDF or Google Slides version separately.