Live signal feed

DeepSeek price change: 1m-output-tokens

Input: $0.28 → $0.28/1M ↑ | Output: $0.877533 → $0.87/1M ↓

DeepSeek Pricing

DeepSeek price change: 1m-input-tokens-cache-miss

Input: $0.14 → $0.14/1M ↑ | Output: $0.4357531 → $0.435/1M ↓

DeepSeek Pricing

Meta price change: llama-3.3-70b

Input: $0.88 → $1.04/1M ↑ | Output: $0 → $null/1M ↑

Meta Pricing

Meta price change: llama-3-8b-instruct-lite

Input: $0.1 → $0.14/1M ↑ | Output: $0 → $null/1M ↑

Meta Pricing

Alibaba price change: qwen3.5-9b

Input: $0.1 → $0.17/1M ↑ | Output: $0.15 → $0.25/1M ↑

Alibaba Pricing

Mistral new model pricing: verified-students-can-get-mistral-pro-for

Mistral Pro student pricing: $6/1M input, $15/1M output. Decent discount if you qualify, but verify eligibility requirements.

Mistral Pricing

YouTube will let you ask AI to make a custom video feed

YouTube adds AI-powered custom feed creation. Interesting UX pattern but no API access for solo devs to leverage yet.

The Verge AI

v0.22.0rc3: [BugFix] Fix hard-coded timeout for multi-API-server startup (#43768)

vLLM fixes timeout bug for multi-API server startup. Minor but useful if you run vLLM.

vLLM releases

Discovery Agents for Real-Time Analytics: Toward Proactive Insight Systems

Multi-agent system for autonomous insight discovery in streaming data. Interesting for real-time analytics apps.

ArXiv cs.AI

GenSBI: Generative Methods for Simulation-Based Inference in JAX

GenSBI: JAX library for simulation-based inference with flow/diffusion models. Practical tool for scientific computing workflows.

ArXiv cs.LG