# Reflection AI to Launch US Open-Weight Model; Updates from OpenAI and Ant Group Reflection AI plans to release its first open-weight model soon, aimed at competing with Chinese models DeepSeek and Qwen. While details like name and… **Published:** 2026-10-05 · **Author:** Yash Thakker **Source:** https://explainx.ai/newsletter/reflection-ai-to-launch-us-open-weight-model-updates-from --- > **In short:** Reflection AI plans to release its first open-weight model soon, aimed at competing with Chinese models DeepSeek and Qwen. While details like name and licensing are pending, businesses should prepare by establishing evaluation tasks and serving paths with current models. ## Reflection AI Set to Release US Open-Weight Model ### Reflection AI Is About to Release a US Open-Weight Model to Take On DeepSeek and Qwen Reflection AI plans to launch its first open-weight model soon, aiming to compete with leading Chinese models like DeepSeek and Qwen. Details on the model's name, size, or licensing are not yet available, but it is expected to initially lag behind top closed models. **What you can do:** For businesses, preparing evaluation tasks and establishing a serving path with current open models can facilitate a smoother transition once Reflection's model is available. [Source](https://explainx.ai/blog/reflection-ai-open-weight-model-us-answer-deepseek-qwen-october-2026) ## AI News Briefs ### OpenAI Promises Codex 28 Days of Daily Improvements or Full Resets OpenAI's Tibo Sottiaux announced a 28-day commitment to deliver daily improvements or full resets for Codex users. This aims to address user complaints about usage limits and enhance overall efficiency, though the exact number of resets is not specified in the pledge. [Source](https://explainx.ai/blog/openai-codex-28-days-daily-improvement-or-reset-tibo-2026) ### Ling-3.1-flash: Ant Group's 560B Open-Weights Model Is Free on OpenCode Ant Group has released Ling-3.1-flash, a free open-weights model with 560 billion total parameters, available on OpenCode. It ranks second among open-weight assistants, but users should note potential limitations in context window and licensing. [Source](https://explainx.ai/blog/ling-3-1-flash-open-weights-free-opencode-2026) ### Claude Now Asks to Train on Your Voice Chats: What the Opt-In Means Anthropic has introduced an opt-in feature for Claude users to allow voice data usage for model improvement. This setting is off by default and separate from text chat training, raising privacy considerations for users. [Source](https://explainx.ai/blog/claude-voice-data-training-opt-in-anthropic-settings-guide-2026) ## Understanding Jev's Speedy Decision-Making ### How Does Jev Actually Work? RLCD and the "System One" Mechanism Jev's unique approach allows it to compute probabilities for multiple answers in one go, skipping the slow, step-by-step token generation of traditional models. This makes it significantly faster and cheaper for structured tasks. **What you can do:** To leverage Jev's speed in your work, consider using it for tasks that require quick, structured decisions, like scoring options or making choices from a limited set. [Source](https://explainx.ai/blog/how-does-jev-work-rlcd-system-one-model-explained-2026) ## Upcoming Workshop on AI Decision-Making ### Build with Jev: The AI That Decides, Not Writes Join a 2-day live workshop on Jev, TypeSafe's System One model — typed Choice, Score and Noul decisions with calibrated confidence, wired into triage, guardrails, routing and browser loops. Led by Yash Thakker. Starts on October 20, 2026; about 2 hours a day. [Source](https://explainx.ai/workshops/jev) ## Reader poll What do you think about the upcoming release of Reflection AI's open-weight model? - Excited to see how it competes - Cautious about its capabilities - Uninterested in open-weight models - Not sure yet ## Sources - [Reflection AI Is About to Release a US Open-Weight Model to Take On DeepSeek and Qwen](https://explainx.ai/blog/reflection-ai-open-weight-model-us-answer-deepseek-qwen-october-2026) - [OpenAI Promises Codex 28 Days of Daily Improvements or Full Resets](https://explainx.ai/blog/openai-codex-28-days-daily-improvement-or-reset-tibo-2026) - [Ling-3.1-flash: Ant Group's 560B Open-Weights Model Is Free on OpenCode](https://explainx.ai/blog/ling-3-1-flash-open-weights-free-opencode-2026) - [Claude Now Asks to Train on Your Voice Chats: What the Opt-In Means](https://explainx.ai/blog/claude-voice-data-training-opt-in-anthropic-settings-guide-2026) - [How Does Jev Actually Work? RLCD and the "System One" Mechanism](https://explainx.ai/blog/how-does-jev-work-rlcd-system-one-model-explained-2026) - [Build with Jev: The AI That Decides, Not Writes](https://explainx.ai/workshops/jev) --- From [explainx.ai](https://explainx.ai/newsletter). Read online, comment and subscribe: https://explainx.ai/newsletter/reflection-ai-to-launch-us-open-weight-model-updates-from