AI Dev Insider

Mistral AI launches Forge, enabling enterprises to train proprietary models on their own data. The new platform includes Mistral Small 4, shifting power dynamics in the AI stack.

Cohere extends its Command A text model with vision capabilities in a new 112B multimodal model. Here's what builders need to know about enterprise document processing at scale.

Mistral AI launches Mistral Small 4 with hardware-efficient design for cost-effective inference. Here's what builders need to know about deployment trade-offs.

SuperAGI's new Inline Voice Agents let you speak commands directly to agents. Here's what this means for your workflow and what you should test.

Together AI's new state space model delivers faster decode speeds than Transformers while staying open-source. What builders need to know about the inference efficiency shift.

Together AI expanded fine-tuning with native tool calling, reasoning, and vision support plus 6× throughput gains. What builders need to know.

Google is discontinuing Firebase Studio less than a year after launch. Here's what this means for your development stack and what you should do immediately.

Windsurf replaces its credit-based model with fixed Free, Pro, Teams, and Max plans. Here's what the shift means for your development workflow and budget planning.

Braintrust adds Agent tool call tracing and experiment parameter management. What builders need to know about improved observability.

v1.660.0 lets developers preview flows using local script changes before pushing to production, cutting iteration cycles for script-heavy workflows.

Windmill v1.661.0 introduces OTel metrics support, giving operators native observability into workflow performance and infrastructure health. Here's what changed and why it matters.

Ollama adds web search and content fetching to OpenClaw, letting local models access real-time data. Local users need authentication to enable the feature.