> ## Content Index
> Fetch the complete content index at: https://www.implicator.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# LLM Meter — Week of May 17
- URL: https://www.implicator.ai/llm-meter-week-of-may-17/
- Published: 2026-05-18T07:17:44.000Z
- Updated: 2026-05-18T07:18:46.000Z
- Author: Marcus Schuler
- Tags: #llm-meter

—GEMINI---  
score: 88  
trend: up  
change: +1  
\+ Snowflake Cortex AI gains native Gemini 3 access — a procurement-frictionless distribution layer into Fortune 500 data warehouses without data egress  
\+ Mars designates Gemini Enterprise as its primary AI operating system across snacking, petcare, and food and nutrition segments  
\+ Gemini Enterprise 3.1 Pro and 3 Flash in Limited Availability with published SLOs — concrete enterprise procurement object  
\+ Google I/O 2026 opens May 19 with new Gemini model expected to anchor the agentic-product narrative for the next cycle  
\- Preview coverage frames the I/O Gemini release as landing at GPT-5.5 level and short of Claude Mythos — sets up post-event score risk  
\---CLAUDE---  
score: 87  
trend: down  
change: -2  
\+ PwC alliance expands with a joint Center of Excellence and 30,000 PwC professionals trained on Claude, Claude Cowork, and Claude Code  
\+ SAP Sapphire embeds Claude as primary reasoning and agentic capability on the SAP Business AI Platform  
\+ $200M Gates Foundation partnership plus Claude for Small Business launch with QuickBooks, PayPal, HubSpot, M365 connectors and a ten-city training tour  
\- June 15 Agent SDK metering puts programmatic Claude on a separate credit pool — 12x to 175x effective price increase for heavy workloads  
\- Plan-limit transparency gap: Anthropic prices the API to the token but refuses to publish the token denominators behind Pro, Max, and Team caps, leaving subscribers unable to verify when limits move  
\---CHATGPT---  
score: 84  
trend: up  
change: +1  
\+ Daybreak launched May 11 — GPT-5.5 and Codex Security packaged for vulnerability detection and patch validation with Cloudflare, Cisco, CrowdStrike, Oracle, Zscaler at launch  
\+ Greg Brockman returns to product chief May 16, merging ChatGPT, Codex, and the developer API into one team to compress execution before Google I/O  
\+ Deployment Company plus Tomoro acquisition brings about 150 Forward Deployed Engineers in-house for enterprise build-outs  
\- Sora and the science division killed in the reorg — narrative cost for a company that has reorganized repeatedly this year  
\- DALL·E 2 and 3 plus Realtime API Beta deprecated May 12 — minor migration friction at the bottom of the stack  
\---MISTRAL---  
score: 73  
trend: down  
change: -1  
\+ Le Chat Work mode and Vibe Remote Agents continue to attract European enterprise pilots on the back of Medium 3.5  
\+ Pre-existing $830M Paris datacenter financing and the Workflows orchestration engine still position Mistral as the EU-sovereign procurement option  
\- Quiet news week with no major customer or product announcement while Anthropic and Google absorbed the enterprise news cycle  
\- Outside the Pentagon IL6/IL7 vendor roster and outside the US-centric agentic-cyber story Daybreak just defined  
\---GROK---  
score: 30  
trend: down  
change: -3  
\+ Wall Street pilot push lands Morgan Stanley and Apollo Global Management as test users alongside other model providers  
\+ SpaceXAI structure inherits SpaceX's federal procurement standing and the GenAI.mil IL5 deployment continues  
\- xAI dissolved into SpaceXAI on May 11 with Musk admitting it was "not built right" — corporate-stability signal that procurement teams hate  
\- Bloomberg reports financiers in the pilot are rarely using Grok for actual work — adoption tell beneath the logo wins  
\- Grok standalone app downloads fell from 20M in January to 8.3M in April per AppMagic — usage curve is collapsing, not stabilizing  
\---DEEPSEEK---  
score: 15  
trend: up  
change: +1  
\+ First external funding round in talks at $45–50B valuation, led by China's Big Fund with Tencent and Alibaba in discussions — closes the undercapitalization argument  
\+ V4-Pro pricing still roughly one-seventh of GPT-5.5 output cost — APAC and self-hosted-research economics intact  
\- V4 launch did not trigger the global reaction of prior releases — competitors have closed the open-weight gap  
\- Big Fund-led financing tightens the perceived Chinese-state link and reinforces the regulatory perimeter in EU, Canada, India, Australia, South Korea