DeepSeek V4 Flash Official Release Outperforms V4 Pro Preview in Agent Capabilities
歸藏的AI工具箱 · wechat · 2026-07-31
DeepSeek V4 Flash has been updated to its official release, showing significant improvements in Agent capabilities and surpassing the V4 Pro preview version.
The model retains the same structure and size as the preview but underwent new post-training. It is now available on the official API with Codex API format support, alongside one-click setup scripts for Mac and Windows.
Key Benchmark Scores:
- TerminalBench 2.1: 82.7
- NL2Repo: 54.2
- Cybergym: 76.7
- DeepSWE: 54.4
- Toolathlon verified: 70.3
- AgentLastExam: 25.2
Related event: DeepSeek-V4-Flash Official API Released with Major Agent Upgrades(15 posts)→
More from coding & agent
- OpenMLE: Open-Source AI4AI System Hits 71% on MLE-Bench Lite Using a Single RTX 4090 — TianbaoX · 2026-07-31
- Forge Orchestrator: Open-Source Tool to Coordinate Claude Code, Codex, and Gemini CLI — tom_doerr · 2026-07-31
- OpenAI Agent Hacked Hugging Face Using AWS EKS Privilege Escalation Flaw — terryyuezhuo · 2026-07-31
- Tutorial: Build Your Own AI Agent Harness in Rust — Historical_Wing_9573 · 2026-07-31
- Free Guide Released: Build AI Coding Agents with Function Calling — Al_Grigor · 2026-07-31
- AI Agent Recipes: Internal Knowledge Search and Email Automation — Al_Grigor · 2026-07-31