The Race for the Next Big Weight-Loss Drug Is On—and China Wants In36m ago DeepSeek launches V4.1-Flash and retires V4-Pro, its flagship model51m ago OpenAI Wants to Know if an AI Industry Slowdown Would Even Be Legal1h ago Oracle Posts Cloud Sales That Top Estimates on AI Demand2h ago iPhone 18 Pro Max vs Samsung Galaxy S26 Ultra: How these XL phones compete2h ago AWS open-sources Pizza Bot: email-style inbox for background AI agents2h ago ToolGrad: Efficient tool-use dataset generation with textual "gradients"2h ago CA governor signs 'landmark' laws on youth use of social media and AI chatbots2h ago OpenAI pulls out of Caltech math hackathon after mathematicians issue open letter criticizing AI math 'slop'2h ago OpenAI spent millions to solve this famous math problem — mathematicians are furious2h ago OpenAI asks Congress for mandatory national AI safety rules before it adjourns2h ago Anthropic Details Disrupted Claude Misuse Across Seven Harm Areas2h ago Meet Redis LangCache: A Managed Semantic Cache That Cuts LLM API Costs by Up to 90% and Returns Cache Hits Up to 15x Faster2h ago Reduce LLM latency with prefix-aware routing on Amazon SageMaker Inference3h ago

The Race for the Next Big Weight-Loss Drug Is On—and China Wants In36m ago DeepSeek launches V4.1-Flash and retires V4-Pro, its flagship model51m ago OpenAI Wants to Know if an AI Industry Slowdown Would Even Be Legal1h ago Oracle Posts Cloud Sales That Top Estimates on AI Demand2h ago iPhone 18 Pro Max vs Samsung Galaxy S26 Ultra: How these XL phones compete2h ago AWS open-sources Pizza Bot: email-style inbox for background AI agents2h ago ToolGrad: Efficient tool-use dataset generation with textual "gradients"2h ago CA governor signs 'landmark' laws on youth use of social media and AI chatbots2h ago OpenAI pulls out of Caltech math hackathon after mathematicians issue open letter criticizing AI math 'slop'2h ago OpenAI spent millions to solve this famous math problem — mathematicians are furious2h ago OpenAI asks Congress for mandatory national AI safety rules before it adjourns2h ago Anthropic Details Disrupted Claude Misuse Across Seven Harm Areas2h ago Meet Redis LangCache: A Managed Semantic Cache That Cuts LLM API Costs by Up to 90% and Returns Cache Hits Up to 15x Faster2h ago Reduce LLM latency with prefix-aware routing on Amazon SageMaker Inference3h ago

Agents1 min reading time

DeepSeek V4 Flash now runs updated weights on AI Gateway

VercelPublished Jul 31, 2026

DeepSeek has updated its V4 Flash model weights on AI Gateway, boosting its agentic performance significantly, with a Terminal-Bench score increase from 56.9 to 82.7. The update is live by default without changing model IDs, and other providers will offer the new weights soon.

#ai-gateway #coding-agents #deepseek-v4-flash #terminal-bench

More in Agents

Meta Announces Muse AI Agent for Personal Tasks and Organization

Covered by 11 sources2d ago

Introducing the Agents API

Covered by 3 sources1d ago

Agents4 min read

Amazon makes its agentic AI platform Quick generally available for desktop on Windows and macOS

Covered by 2 sources1d ago