AI Coding Live News

On Nov 2, semiconductor and AI research firm SemiAnalysis released GDPx, a benchmark of agentic capability across 11 domains: chip design, cybersecurity, biology, finance, legal, quant trading, recursive self-improvement (RSI), hardware design, materials science, alignment and software engineering. Its 206 tasks were built by 14 specialist data partners, with SemiAnalysis doing final QA. On the official leaderboard Claude Opus 5.5 leads at 12.0%, followed by Gemini 4 Carbon at 10.9% and GPT-6 Astra at 9.6%; nearly every model scores 0 on finance and legal. SemiAnalysis says safety classifiers drag down Opus 5.5 and GPT-6 Astra, and plans a new version every 3–6 months. All scores are as reported by SemiAnalysis.

Zoom

On Oct 29 Anthropic introduced self-assessment on X: after each response, Claude produces a structured preliminary safety assessment of the whole conversation for offline review. Anthropic's Safeguards blog says that on Claude Fable 5, at a 1% false-positive rate, it flags 78% of the conversations a stronger prompted classifier flags and 94% of high-priority ones. Because the conversation is already in the KV cache, it adds about 10 output tokens on average and an estimated under 2% of inference compute. It has run in production since June, but Anthropic says it is not robust against an attacker who targets it directly and is only one layer of defense in depth.

Zoom

BerriAI shipped LiteLLM v1.103.0 on 2026-09-28: delegated MCP OAuth now requires admission; Capability classifier + Fuse V2 routers land; the proxy can predict prompt-cache costs across deployments, bind JWT claims to agents, add tpd_limit, bulk user/team APIs, and /nvidia_nim passthrough. Also: Bedrock S3 managed file delete/list, Friendli price auto-sync, Responses↔Chat reasoning mapping, and cosign image verification docs.

Zoom

Confident AI’s deepeval 4.2.x (notes for 4.2.2/4.2.4; PyPI at 4.2.6) wires TypeSafe Jev into LLM-as-a-judge: evaluation LLMs still handle language tasks, while closed verdicts go to System One for typed decisions with calibrated probabilities. Claimed decision-stage pricing is about $0.042 per million input tokens with no output charge, ~100ms typical latency, and batched independent questions. Faithfulness, GEval, DAGMetric, and Classifiers are integrated; enable via deepeval set-mode experimental plus typesafe-sdk—missing keys fail loudly instead of silent fallback.

Zoom

Safety researchers have introduced 'Just Ask Jev' alongside the RLCDAlignBench suite (arXiv: 2609.29429, GitHub: sumleo/RLCDAlignBench). Conventional alignment auditors rely on expensive generative LLM judges or rigid single-label classifiers (e.g., Llama Guard). Trained via Reinforcement Learning for Calibrated Decisions (RLCD), Jev evaluates multiple typed alignment questions simultaneously across ten failure modes—sycophancy, jailbreaks, deception, prompt injection, hallucination, privacy breaches, bias, reward hacking, uncertainty concealment, and power seeking—in a single forward pass. Achieving a median 0.886 zero-shot AUROC across 44 benchmarks, Jev matches human consensus while slashing evaluation costs by 63x.

Zoom

BerriAI shipped LiteLLM v1.102.1 on 2026-09-23 (PyPI litellm 1.102.1). Versus 1.102.0 it backports Anthropic/Bedrock/Vertex safeguards→safeguard_results passthrough so Claude Code Auto mode through a LiteLLM gateway can show Auto mode server: Enabled, and cherry-picks seven TypeSafe Jev changes (passthrough, complexity router, relevance compaction guardrail, auto-router classifier, OpenRouter jev-1.13 pricing, etc.) from the 1.103 RC onto stable/1.102.x. Docker images keep cosign verify instructions.

Zoom

Pipecat v1.12.0 (2026-09-26) adds TTSService.pronunciation_transform_ipa() for reusable word→IPA maps across Cartesia/ElevenLabs/Inworld/Deepgram, client-mode MOQTransport reconnect with session-ending markers, Pipecat Classifiers + LLMClassifier, frame-level interruptible, and response_schema on LLM run_inference. PyPI pipecat-ai 1.12.0 is live.

Zoom

Mastra shipped @mastra/[email protected] on 2026-09-24: register classifiers on new Mastra({ classifiers }) and use them as typed workflow steps; ClassifierProcessor fails closed by default on I/O/stream; tools can context.background.adopt() long work; @mastra/connect adds ten generated SaaS tool providers (Slack, GitHub, Gmail, …).

Zoom