OrcaRouter 🐳

@OrcaRouter

The AI Gateway built for Privacy & Security 🐳 0% Markup. BYOK. Adaptive Routing. Firewall. Guardrails. Observability. Cyber. → discord.gg/YEubt8enRA

orcarouter.ai

Joined April 2026

Replies Reposts Media Articles

Pinned
OrcaRouter 🐳
@OrcaRouter
Sep 29
Anthropic just sounded the alarm on GLM-5.3’s cyber capabilities. We built on it. OrcaCyber Zero 1.0 is based on the GLM-5.3, post-trained specifically for vulnerability research, exploit reproduction, red teaming, and autonomous cyber workflows. 98.01% pass@1 on CyberGym L1.
32
OrcaRouter 🐳
@OrcaRouter
7h
Following our post, we just analyzed and open sourced the system prompts behind 12 open + closed-source coding agent harnesses — Claude Code, Codex, Cursor, Devin, Factory Droid, OpenCode, Cline, Kilo, MiMoCode, MiniMax Code, GLM ZCode & DeepSeek DSH. We looked at two things: →
OrcaRouter 🐳
@OrcaRouter
7h
If the secret sauce is prompts, context engineering, tool loops, routing, and evals, how long before the moat gets reverse-engineered?
16
OrcaRouter 🐳
@OrcaRouter
7h
If the secret sauce is prompts, context engineering, tool loops, routing, and evals, how long before the moat gets reverse-engineered?
Matan Grinberg
@matanSF
9h
We are terminating Chris Degnan for unethical conduct involving Cognition. The last few months have seen incredible progress in AI capabilities. San Francisco has flourished as new companies that solve new, more ambitious problems are finding great success. Generally, it is a
2
OrcaRouter 🐳
@OrcaRouter
Sep 29
Why are we making AI agents rediscover the same relationships on every query? Our new paper, IndexRAG, accepted to Findings of AACL-IJCNLP 2026, explores a different approach: move part of the reasoning to index time. Instead of retrieving documents and reasoning over them
arxiv.org
IndexRAG: Bridging Facts for Cross-Document Reasoning at Index Time
Multi-hop question answering (QA) requires reasoning across multiple documents, yet existing retrieval-augmented generation (RAG) approaches address this either through graph-based methods...
2
OrcaRouter 🐳
@OrcaRouter
Sep 29
Great independent local-model benchmark on an RTX 3090. OrcaSAQ-2-27B: • 90.8% MATH-500 — best local result • 55.0% GPQA-D — best local result • 89.6% HumanEval+ • 262K context • ~12 GB footprint (among the smallest) What matters here is the tradeoff: capability per GB,
Justin Brock (e/acc)
@justinbrock
Sep 29
Just tested OrcaSAQ-2-27B on a single RTX 3090 (24GB VRAM). Hybrid attention (48 Gated DeltaNet + 16 full-attention layers) cut KV cache memory ~4x, fitting 262K context with 6.5GB headroom. Best local MATH-500 at 90.8%. "Thinking mode" breaks standard evals.
6