Google's Gemini moved ahead of OpenAI's ChatGPT this week in Implicator's LLM Meter, the first crossover since the weekly enterprise scorecard launched in March. Gemini gained two points to 81 on the strength of its price advantage over Claude Opus 4.7 and the imminent Google Cloud Next '26 keynote. ChatGPT lost two points to 79 after OpenAI shed three senior leaders and saw its enterprise AI spend share drop from 50% to 27%.

Key Takeaways

AI-generated summary, reviewed by an editor. More on our AI guidelines.

Claude held the top slot at 88, down one. Grok fell five points to 35. DeepSeek ticked down one to 13. Mistral rose two points to 69.

Gemini's rise runs on price and timing

Gemini 3.1 Pro is now available in preview across Vertex AI and Gemini Enterprise at $2 per million input tokens and $12 per million output tokens. That runs 2.14 times cheaper than Opus 4.7 on base pricing, and SWE-bench Verified scores sit within two-tenths of a point of Opus 4.6. Price, not capability, is the spread.

Google Cloud Next '26 opens April 22 in Las Vegas with an agenda packed around agentic workflows, governance, and security, plus third-party agents from ServiceNow and Comeen. A Workspace reading suggests Gemini now handles roughly 45% of Fortune 500 enterprise AI queries inside Workspace deployments. Dropbox federated data store went GA. Mariner Computer Use shipped. That is a lot of enterprise plumbing landing in one week.

Claude retakes coding. Operations keep pulling it back.

Anthropic shipped Claude Opus 4.7 on April 16. SWE-bench Verified climbed from 80.8% to 87.6%, an 11-point jump in one version bump. SWE-bench Pro did even better, moving 53.4% to 64.3%. The model beats GPT-5.4 and Gemini 3.1 Pro on directly comparable coding benchmarks. Enterprise momentum is real: Ramp data has Anthropic at 30.6% of AI-paying business customers against OpenAI's 35.2%, with a projected crossover within two months.

The operational side told a different story. Claude suffered three disruptions between April 10 and April 15, with 90-day API uptime at 98.95% against the 99.99% enterprise standard. Anthropic also unbundled Claude Code from enterprise seat fees, moving heavy users to per-token billing. Retool CEO David Hsu switched his company's coding workloads to OpenAI over reliability.

Grok's five-point slide came from three directions

NBC News reported April 14 that Apple privately threatened to pull Grok from the App Store over non-consensual sexualized deepfakes. On April 9, xAI sued Colorado in federal court to block the state's AI anti-discrimination law, arguing it would force Grok to abandon its "pursuit of truth." And SpaceX's confidential April 1 filing for a June Nasdaq listing at up to $1.75T comes with a mandatory Grok subscription requirement for banks, law firms, and auditors bidding on IPO roles.

Three compliance red flags in one week. Procurement teams read that kind of pattern and move on.

What to watch next week

Google Cloud Next '26 runs April 22-24. Expect movement on the Gemini line based on what ships on stage. DeepSeek's V4 is reportedly due in late April. Anthropic's civilian Mythos lane out of the White House meeting is still under negotiation. Opus 4.7 could hold coding through the summer. Nothing else on the board looks settled.

Frequently Asked Questions

What is the Implicator LLM Meter?

It's a weekly enterprise scorecard. Six models sit on the board: Claude, ChatGPT, Gemini, Mistral, Grok, DeepSeek. Each score runs 0-100, weighted toward compliance posture and reliability above raw model quality. Pricing matters. So does vendor stability. Trends move week over week based on what actually shipped.

Why did Gemini pass ChatGPT this week?

Two factors. Gemini 3.1 Pro's base pricing runs roughly 2.14 times cheaper than Claude Opus 4.7 at comparable SWE-bench scores, giving it the value lead for high-volume enterprise workloads. OpenAI simultaneously lost three senior leaders and watched its enterprise AI spend share fall from 50% to 27% over the past year.

What caused Grok's five-point drop?

Three compliance red flags in the same week. Apple privately threatened to remove Grok from the App Store over non-consensual sexualized deepfakes. xAI sued Colorado in federal court to block the state's AI anti-discrimination law. And SpaceX required banks and law firms bidding on its June IPO to purchase Grok subscriptions as a condition of participation.

Why did Claude lose a point despite winning Opus 4.7?

Opus 4.7 retook the coding lead with SWE-bench Verified at 87.6%, but operational problems weighed the score down. Claude had three disruptions between April 10 and April 15, with 90-day API uptime at 98.95% against the 99.99% enterprise standard. Anthropic also unbundled Claude Code from enterprise seat fees, raising bills for heavy users.

What should enterprise buyers watch next week?

Google Cloud Next '26 runs April 22-24 with major agentic, governance, and security announcements expected. DeepSeek's long-delayed V4 is reportedly due in late April. And the civilian Mythos Preview lane from Anthropic's White House meeting with Wiles and Bessent is still under negotiation with civilian agencies.

AI-generated summary, reviewed by an editor. More on our AI guidelines.

AI News

Los Angeles

Tech culture and generative AI reporter covering the intersection of AI with digital culture, consumer behavior, and content creation platforms. Focusing on technology's beneficiaries and those left behind by AI adoption. Based in California.