Alon Amit (@alonamit), a VP at Intuit who posts about math, wrote on X at 07:46 SGT on November 15 that he asked Gemini 4 Carbon a math question and unintentionally left the Thinking level on Low; it started answering immediately, and after six screenfuls of math the screenshot he shared appeared. In it the model talks to itself: it had closed </thought> at the start and never reopened it, "I WAS WRITING DIRECTLY TO THE USER?!", blaming "effort level is 0.05 so my initial thought block closed immediately," noting it cannot delete already-emitted tokens, and telling itself "Okay, stop and breathe." Amit said it "went on like this for pages and pages." The post has about 3,291 likes, 99 reposts, 62 quotes, 149 replies and roughly 223K views, and many accounts reposted it as a joke. This is one user's experience; we found no public response from Google.
Key Takeaways
- ✓The post: Alon Amit says he unintentionally left Thinking on Low, Carbon answered immediately, and after six screenfuls of math it started talking to itself and "went on like this for pages and pages," with about 3,291 likes, 862 bookmarks and roughly 223K views (post)
- ✓The model's own explanation in the screenshot: it closed </thought> at the start and never reopened it, and "effort level is 0.05" made its initial thought block close immediately, so the scratchpad went straight into the user-facing answer
- ✓It then noted it "cannot delete already-emitted tokens," told itself to "stop and breathe" and pivot to a clean answer, yet kept writing out its reasoning
- ✓The debate: Low thinking saves time and cost, but this case suggests the line between thinking and answering can slip at low effort; many joked it is "the most human AI," while some said they would not put Low mode in front of customers
- ✓This is one user's screenshot and opinion, not evidence of how Carbon usually behaves; we found no public response from Google, so treat it as discussion
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background
Gemini 4 Carbon is the flagship model Google released in late October, with adjustable thinking levels. Low is fast and cheap and is often used for simple questions. On November 15, Alon Amit (a VP at Intuit) shared on X a case where Low went wrong: the model wrote its scratchpad straight into the answer, then noticed.
What happened under the hood
By the model's own account in the screenshot, it emitted the closing </thought> tag at the very start and never reopened a thought block, so all later reasoning went into the user-visible output stream; it blamed "effort level is 0.05." That is only the model's self-explanation; Google has not published how the Low level works internally.
Numbers that matter
This is not a benchmark story and there are no model scores. The only verifiable figures are post metrics: about 3,291 likes, 99 reposts, 62 quotes, 149 replies and roughly 223K views.
What developers should take from it
When using a low thinking level via the API or an app, do not assume the model will keep reasoning and answer cleanly apart: for user-facing output, add length and format checks, use a higher level for hard math or multi-step reasoning, and test each level on your own samples before shipping.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.