AICoder

AI Coding Live News

First-hand AI coding updates from 𝕏/Twitter frontier labs and researchers, distilled by AI with entity benchmarks.

Explore live on 𝕏 ↗

AI Coding Live News

M
ModelScope@ModelScope2022·
👀133
🔥85
View on 𝕏 ↗

Shanghai AI Lab ships Atria Dawn full release: 744B MoE GLM-5.2 agentic model, context doubled to 512K, MIT weights

Shanghai AI Laboratory has moved Atria Dawn from Preview to a full release, with open weights (plus an FP8 build) on Hugging Face under internlm/Atria-Dawn and on ModelScope, MIT licensed. Built on the 744B-parameter MoE GLM-5.2, it targets long-horizon research and engineering agent work; context doubles from the Preview's 256K to 512K and input is text only. The vendor table shows it leading the listed models on AutomationBench, CyberGym and SkillsBench, while still trailing Claude Opus 5 on coding tests such as SWE-bench Pro and Terminal-Bench 2.1 (all vendor-reported).

Shanghai AI Lab ships Atria Dawn full release: 744B MoE GLM-5.2 agentic model, context doubled to 512K, MIT weights

Key Takeaways

  • 512K context, double the Preview's 256K; built on the 744B-parameter MoE GLM-5.2, text input only
  • BF16 and FP8 weights on Hugging Face and ModelScope, MIT license
  • Vendor-reported: AutomationBench 54.7, CyberGym 86.8, SkillsBench 67.8, BFCL V4 78.4, each the top score in its table
  • Weak spots from the same table: SWE-bench Pro 61.7 vs 76.5 for Claude Opus 5, Terminal-Bench 2.1 81.0 vs 90.2
  • Local serving needs SGLang v0.5.13.post1+ or vLLM v0.23.0+; an OpenAI-compatible API also plugs into Codex
Read full story
Linked Models & Agents:
⚡ Atria Dawn⚡ Atria Dawn Preview⚡ GLM-5.2🤖 Codex🌐 Shanghai AI Laboratory🌐 ModelScope🌐 Hugging Face
#Atria Dawn#Shanghai AI Lab#open weights#agentic model#MoE#512K context#MIT
@
@ollama@ollama·
👀51
🔥94
View on 𝕏 ↗

Ollama Cloud Launches US/EU Zero-Data-Retention Hosting for DeepSeek-V4.1-Flash-1017 with Claude Code CLI, While SemiAnalysis Reveals GLM-5.3 Was Built 100% via Post-Training

Ollama launched US- and EU-hosted Ollama Cloud with strict zero data retention and zero training, adding immediate support for DeepSeek-V4.1-Flash-1017 via one-line launch commands for Claude Code and OpenCode alongside Ollama v0.40.6. Meanwhile, SemiAnalysis revealed that GLM-5.3 was built entirely by post-training the previous GLM-5.2 base model without pretraining from scratch, achieving a 'night and day difference' in cybersecurity and agentic capabilities.

Ollama Cloud Launches US/EU Zero-Data-Retention Hosting for DeepSeek-V4.1-Flash-1017 with Claude Code CLI, While SemiAnalysis Reveals GLM-5.3 Was Built 100% via Post-Training

Key Takeaways

  • Ollama Cloud launches US- and EU-hosted zero-data-retention serving for DeepSeek-V4.1-Flash-1017 with one-line ollama launch integrations for Claude Code and OpenCode
  • Ollama v0.40.6 adds GGUF q2_0 tensor support, clef cloud integration, and fixes orphaned runner processes on Windows
  • SemiAnalysis reveals GLM-5.3 was built entirely by post-training the previous model without pretraining from scratch, delivering a 'night and day difference' in cyber capabilities over GLM-5.2
Read full story
Linked Models & Agents:
#Ollama#Claude-Code#OpenCode#DeepSeek-V4.1-Flash#GLM-5.3#SemiAnalysis
A
Anthropic@AnthropicAI·
👀75
🔥91

Anthropic research: small models may be months from weight-based self-replication without a datacenter; IoT fleets can serve 26B shards

On 2026-10-12 Anthropic (William Zhou, Daniel Freeman) published research arguing small models may be months from autonomous weight-based replication on commercial IoT. Models ≤30B already complete many replication steps; in simplified evals GLM-5.2 finishes end-to-end. A demo ran Gemma4-26B-A4B across 15 ~$55 Android TV boxes via pipeline parallel (~0.39 tok/s). Extrapolating Epoch Capabilities Index gains, small models may hit the replication floor by end of 2026.

Key Takeaways

  • Official: Anthropic research post (2026-10-12)
  • Demo: 15 cheap Android TV boxes (~4GB RAM) pipeline-parallel serve Gemma4-26B-A4B at ~0.39 tok/s; a 9B model stood up inference on 8/9 hardware types
  • E2E: in a firewalled redroid VPC with unauthenticated ADB, GLM-5.2 reached 8/8 nodes in 4/5 trials (7/8 once), tasked to serve Llama3.1-8B then spawn
  • Timescale (vendor): full inference-setup trajectories ~71–150h (Gemma4-26B-A4B) / 28–51h (Qwen3.5-9B)—day-scale doubling, not minute-scale worms
  • Defenses: drive R₀ < 1; ISP flow anomalies, honeypots, quarantine; manufacturer defaults and compliance
Read full story
Linked Models & Agents:
⚡ GLM-5.2⚡ Gemma4-26B-A4B⚡ Qwen3.5-9B⚡ Llama3.1-8B🤖 Pi🤖 Memento🌐 Anthropic
#Anthropic#AI Safety#IoT#Self-replication#GLM-5.2#Security#ARA
I
InternLM / Shanghai AI Lab@InternLM·
👀85
🔥91

LMDeploy 0.18.0: Qwen3.5 dflash, SM90 quantized GEMMs, Ascend GLM-5.2; TurboMind W4A16 + piecewise CUDA Graph

Shanghai AI Lab’s LMDeploy shipped v0.18.0 on 2026-09-28: input logprobs, expanded SM90 quantized GEMMs with unified TurboMind linear paths, Ascend GLM-5.2, Qwen3.5 dflash, and MoE shared-expert/FFN sharding. PyTorch path prototypes TurboMind W4A16 (AWQ), piecewise CUDA Graph prefill, XTuner TileLang sparse MLA, checkpoint-engine weight updates, and request-only KV cache metrics.

LMDeploy 0.18.0: Qwen3.5 dflash, SM90 quantized GEMMs, Ascend GLM-5.2; TurboMind W4A16 + piecewise CUDA Graph

Key Takeaways

  • Shipped v0.18.0 — Qwen3.5 dflash, Ascend GLM-5.2, SM90 quantized GEMMs
  • TurboMind W4A16 AWQ prototype + piecewise CUDA Graph prefill + TileLang sparse MLA
  • checkpoint-engine weight updates + request-only KV cache metrics + input logprobs
  • Upgrade: pip install -U "lmdeploy>=0.18.0"
Read full story
Linked Models & Agents:
⚡ Qwen3.5⚡ GLM-5.2🌐 InternLM🌐 Shanghai-AI-Lab
#LMDeploy#TurboMind#Qwen3.5#Dflash#AWQ#CUDA Graph#Ascend#Inference
ADSponsored
A
Atria ASI@AtriaASI·
👀185
🔥95
View on 𝕏 ↗

Shanghai AI Lab open-sources Atria Dawn Preview: long-horizon agentic model on 744B MoE GLM-5.2, MIT

Shanghai AI Lab released Atria Dawn Preview, an agentic foundation model for long-horizon research/engineering tasks built on the 744B-parameter MoE GLM-5.2, with tool use and executable verification loops. Weights are on Hugging Face (internlm) and ModelScope, plus an FP8 build, 256K context, GitHub, website, and API try links.

Key Takeaways

  • Atria Dawn Preview is built on 744B MoE GLM-5.2 for long-horizon research and engineering agents.
  • Open weights on Hugging Face/ModelScope with FP8 and 256K context under MIT.
  • Covers Discovery/Creation/Delivery/Cybersecurity end-to-end scenarios with API try access.
Read full story
Linked Models & Agents:
⚡ GLM-5.2🌐 shanghai-ai-lab
#Atria Dawn Preview#Shanghai AI Lab#GLM-5.2#744B MoE#open weights#MIT#Agentic#256K
T
Together AI@togethercompute·
👀65
🔥91
View on 𝕏 ↗

Together: GLM-5.3 beats GPT-5.6 Sol and Claude Fable 5 on agentic benchmarks

Together AI said GLM-5.3 now beats GPT-5.6 Sol and Claude Fable 5 on agentic benchmarks, with GLM-5.3 Flash close behind. Zhipu kept the GLM-5.2 base and scaled post-training—more long-horizon environments, more diverse tasks, and more RL compute—rather than training a new foundation model. For coding agents, that is a claim that post-training stack height can overtake closed flagships, and that the cheaper open-weight default is already on Together’s API.

Key Takeaways

  • Together reports GLM-5.3 ahead of GPT-5.6 Sol and Claude Fable 5 on agentic benchmarks; Flash is next.
  • The gain is post-training on the GLM-5.2 base: more long-horizon RL, not a new pretrained model.
  • Coding agents can switch to a cheaper open-weight default without waiting on a new foundation run.
Read full story
Linked Models & Agents:
⚡ GLM-5.3⚡ GPT-5.6 Sol⚡ Claude Fable 5🌐 Zhipu AI (智谱 AI)🌐 OpenAI🌐 Anthropic
#GLM-5.3#Together AI#GPT-5.6 Sol#Claude Fable 5#agentic benchmarks
B
Baseten@baseten·
👀34
🔥88
View on 𝕏 ↗

Baseten Launches GLM-5.3-Flash on Model APIs with 122+ TPS and Zero Data Retention by Default

Baseten launched GLM-5.3-Flash on its Model APIs platform on day zero, achieving 122+ TPS with default Zero Data Retention (ZDR) and US-only infrastructure at a 90% cost reduction over GLM-5.2.

Key Takeaways

  • Achieves top-tier 122+ TPS decode throughput across Artificial Analysis and OpenRouter independent benchmarks;
  • Enterprise compliance guaranteed via US-only isolated compute clusters and strict Zero Data Retention (ZDR);
  • 1M context and native multimodal vision make it a prime cost-effective engine for enterprise coding agents.
Read full story
Linked Models & Agents:
⚡ GLM-5.3-Flash
You've caught up with all stories