2026-06-06

The Week the OS Became the Agent

5 fresh signals — Microsoft Build reframes Windows, Anthropic splits the frontier, OpenClaw hits its security wall, Kimi K2.6 leads open weights, and humanoids leave the demo floor

Live Pulse — 2026-06-06
7
Microsoft MAI Models
88.6%
Opus 4.8 SWE-bench Verif.
77.8%
Mythos Preview SWE-bench Pro
1,184
Malicious ClawHub Skills
42K
Exposed MCP Servers
377K
OpenClaw Stars
1T / 32B
Kimi K2.6 MoE / Active
Story 01 — OS Shift

Microsoft Build 2026 Turns Windows Into an Agent Operating System

Seven in-house MAI models shipped at Build — including MAI-Thinking-1 (matches Opus 4.6 on SWE-bench Pro) and MAI-Code-1-Flash for Copilot. New Project Solara is a chip-to-cloud agent platform. Windows AI APIs moved from NPU-only to CPU + GPU. Microsoft Execution Containers (MXC) sandbox agent blast radius. Files "survived because the OS enforced the rule, not because the model behaved." Microsoft is no longer an AI customer — it's a builder.

Platform Shift Microsoft Build 2026
7
In-house MAI models shipped in one keynote
Story 02 — Frontier

Claude Opus 4.8 + Mythos Preview Split the Frontier — 90-Day Clock Starts

Opus 4.8 (May 28) hits 88.6% on SWE-bench Verified and 69.2% on SWE-bench Pro. Mythos Preview (restricted to Glasswing) takes the Pro crown at 77.8% — 9 points ahead. New capability: dynamic workflows that spawn hundreds of parallel sub-agents per call. Mythos priced at 25/125 dollars per M tokens (5x Opus). Mythos GA in weeks. Every coding-agent startup priced on Opus 4.x economics has a 90-day clock.

Model Race Anthropic Coding Agents
77.8%
Mythos Preview — SWE-bench Pro leader
Story 03 — Security

OpenClaw Hits Industrial-Scale Security Crisis — 1,184 Malicious Skills Confirmed

Antiy CERT confirms 1,184 malicious skills on ClawHub (22% contamination). Trend Micro finds 492 MCP servers exposed with zero authentication. Six CVEs in 2026. 42,000 exposed servers. Pentagon designated Anthropic a "supply-chain risk" — first American company classified that way. Yet OpenClaw gained 186 stars yesterday (now 377,144). The 2015-npm moment for agent skills is here, with the agent's full permission scope as the attack surface.

Security Crisis OpenClaw Supply Chain
22%
Of ClawHub skills confirmed malicious
Story 04 — Open Weights

Kimi K2.6 Becomes the Open-Weight Agent Swarm to Beat

Moonshot's 1T MoE / 32B active model, 256K context, 300 parallel sub-agents, modified MIT license, fits on 4x H100 INT4. Top open-weight on SWE-Bench Pro. Qwen 3.6 27B (April 16) is the consumer-hardware winner at 77.2% SWE-bench in 24GB Q4. The open-weight race is now Kimi K2.6 (frontier) vs Qwen 3.6 27B (consumer) — Mythos-vs-K2.6 at the top tier.

Open Weights Moonshot Apache-like
300
Parallel sub-agents per K2.6 call
Story 05 — Robotics

Humanoid Race Leaves Demo Floor — Figure 03 Has BMW, Tesla Has Nothing

Tesla killed Model S and X (Jan 28) to fund Optimus. Musk admitted on Q4 2025 earnings Optimus is "not in usage in our factories in a material way." Figure 03 is the only humanoid with a paying enterprise customer — BMW Plant Leipzig, plus a 39B dollar valuation. NVIDIA's Physical AI stack (Isaac + Omniverse + Jetson Thor) is the platform layer. The race is over for demos; it's deployed units now.

Deployment Wins Humanoid Figure 03
$39B
Figure valuation — only humanoid with paying customer
SWE-bench Pro — Frontier Model Leaderboard

Coding-agent evaluation — percentage of real-world GitHub issues solved

Mythos Preview
77.8%
Opus 4.8
69.2%
Kimi K2.6
67.1%*
MAI-Thinking-1
57.5%
GPT-5.5
48.2%

*Kimi K2.6 estimated from SWE-Bench Pro open-weight leaderboard. Mythos Preview restricted to Project Glasswing.

OpenClaw Contamination — 22% of Skills Flagged

Antiy CERT audit of ClawHub registry as of June 2026

22%
flagged
Malicious (1,184)
Clean (4,216)
5,400+
Total skills on ClawHub registry
Pentagon designated Anthropic a "supply-chain risk" — first US company
MCP Server Exposure — Internet-Facing, Zero Auth

Trend Micro scan of publicly reachable MCP endpoints

No auth
492
Auth weak
246
Auth strong
123

Snyk for ClawHub/MCP doesn't exist yet. The trust layer is the biggest unfilled gap in agent infrastructure.

Anthropic Frontier Cadence — 9-Point Lead in 60 Days

SWE-bench Pro scores across 2026 releases

April 2026
Opus 4.7 — 64.3%
May 28, 2026
Opus 4.8 — 69.2% (+4.9)
May 2026
Mythos Preview — 77.8% (+8.6)
"Weeks away"
Mythos GA — pricing reset imminent

Threat: Coding-agent startups priced on Opus 4.x economics have a 90-day clock. Dynamic workflows × Mythos price collapse resets the cost curve.

Microsoft Build 2026 — The New Agent Stack

Microsoft is now a model builder, not a customer. Seven proprietary MAI models plus Project Solara agent platform.

Layer
Component
What it does
Status
Models
MAI-Thinking-1, MAI-Code-1-Flash + 5 more
First Microsoft reasoning model, Copilot integration
Foundry preview
Platform
Project Solara
Chip-to-cloud agent platform, "open multiple-agent world"
Announced
Runtime
Windows AI APIs
Now on CPU + GPU, not just NPU
Live
Sandbox
Microsoft Execution Containers (MXC)
Agent blast-radius containment at OS level
Live
Distribution
Microsoft 365 + Dynamics + Azure AI
Default MAI model on every Microsoft surface
Shipping

Key insight: Files "survived because the OS enforced the rule, not because the model behaved." The OS is the new agent primitive. TTL should treat Microsoft as a model competitor, not a customer.

Open-Weight Agent Models — The 2026 Field

Five open-weight models, ranging from consumer laptops to multi-GPU server rigs

Model
Size
Hardware
Headline
Kimi K2.6
1T / 32B active
4x H100 INT4
Top open-weight SWE-Pro
Qwen 3.6 27B
27B
24GB Q4 (laptop)
Best consumer model
GLM-4.7
Q1 2026
Server
HLE ~42.8% tool use
DeepSeek V4 Pro
V4
Server
0.44/M tokens
gpt-oss:20b
20B
Edge / laptop
OpenAI open-weight

Key insight: Kimi K2.6 owns the frontier open-weight slot, Qwen 3.6 27B owns consumer hardware. Mythos-vs-K2.6 is the closed-vs-open frontier race for the next 90 days.

TTL Action Items — This Week
P0 — INFRASTRUCTURE
Resolve RTX offline state
rtx_mac.txt missing — WoL packet going to placeholder MAC. Add real MAC or set up Tailscale + smart plug watchdog. Owner: Sergio, EOW.
P1 — AGENT FLEET
Build Mythos-ready agent fleet
Dynamic workflows: N parallel sub-agents per Hermes call. Target Kimi K2.6 for open-weight, Opus 4.8 for proprietary. Owner: Forge + Kai, 2 weeks.
P1 — SECURITY
Add ClawHub-audit to skill install
Wrap hermes skills install with verification against Antiy CERT + joylarkin security feeds. 1,184 malicious skills is enough reason. Owner: Dragon, 1 week.
Humanoid Robots — Demos vs Deployed Units
Figure 03
$39B
Valuation — paying customer

BMW Plant Leipzig is the only enterprise deployment in production. NVIDIA Isaac + Omniverse + Jetson Thor stack is the platform.

Tesla Optimus
"Not material"
Musk, Q4 2025 earnings

80% of Tesla strategic focus, but "not in usage in our factories in a material way." Model S and X killed Jan 28 to fund it.

Physical AI Stack
Cosmos 3
NVIDIA — open weights

First open-weights physical-AI foundation model with native action modality. Shifts robotics pretraining bottleneck from data to fine-tuning compute.