Gemini 3 Pro: what developers actually say

We read every Hacker News comment from Nov 2025 – Nov 2026 that gave a real opinion on Gemini 3 Pro. 34 of the 42 opinions come from people who say they use it themselves.

52%approval, all opinions

–approval, last 90 days

#41of 45 models ranked

40different developers

Developers give Gemini 3 Pro a 52% approval score across 42 opinions, with 34 from firsthand use. Code quality draws the most positive feedback: 14 positive and 6 negative mentions. Reliability is the clearest weakness: 0 positive and 9 negative mentions, with reports of overload errors, rate limits, and hangs. Approval peaked at 67% in January 2026, fell to 43% in February, and recent samples are thin: 2 opinions in April at 50% and 1 in May at 0%.

Pick it if

Want a second opinion on a large codebase
Need a fallback when Codex limits are hit
Will make it write tests to check its output

Look elsewhere if

Need dependable capacity for long sessions
Need strong tool calling or agentic work
Need a model that is easy to steer

Strengths and weaknesses

Each row counts opinions that mentioned the topic. Bars to the right of the line are praise, bars to the left are complaints.

PraiseComplaints

Code quality

Reliability

Speed

How opinion has moved

Monthly approval (line) and number of opinions (bars). Months with fewer than 3 opinions are left out.

What developers say

Praise

Says Gemini 3 Pro helped narrow down a hard bug when given the full large codebase context.HN comment · Feb 20, 2026

Uses Gemini 3 Pro conversationally for coding help; calls it adequate and a speed boost for rubber-ducking.HN comment · Feb 20, 2026

Used Gemini 3 Pro for codebase reviews, and its hint eventually helped find a hard-to-spot resource leak.HN comment · Jan 26, 2026

Has mostly settled on Gemini 3 Pro and cross-checks code reviews across models to catch blind spots.HN comment · Jan 26, 2026

Finds Gemini 3 Pro very good and useful as a fallback when Codex limits are hit, though slightly behind Codex.HN comment · Jan 15, 2026

Says Gemini 3 Pro works well for coding when the agent writes tests to verify its output.HN comment · Jan 8, 2026

Often prefers the UIs Gemini 3 Pro generates over Opus output.HN comment · Jan 8, 2026

Admits Gemini 3 Pro is not the fastest model but is happy with its high-quality output.HN comment · Jan 8, 2026

Complaints

Says Gemini 3 Pro failed to complete a planned OpenCode refactor over eight days, so they cancelled.HN comment · May 10, 2026

Gemini 3 Pro was great at first but has since gotten worse than smaller models.HN comment · Feb 19, 2026

Finds Gemini 3 Pro not great, often random and prone to going off the rails.HN comment · Feb 2, 2026

Found Gemini 3 Pro less usable and reliable than Kimi K2 Thinking for their specific work.HN comment · Jan 29, 2026

Finds Gemini 3 Pro a step below Claude and GPT, though fast and cheap.HN comment · Jan 26, 2026

Says Gemini 3 Pro has the same speed problem as Codex, plus a much bigger tool-calling gap.HN comment · Jan 23, 2026

Reports Gemini 3 Pro frequently returns overload errors, seemingly due to capacity limits.HN comment · Jan 21, 2026

Says Gemini hits rate limit errors and occasionally hangs, so its infrastructure feels unreliable.HN comment · Jan 8, 2026

Summaries are paraphrased. Each links to the original comment.

Gemini 3 Pro compared

Gemini 3 Pro vs Claude Opus 4.5

FAQ

Is Gemini 3 Pro worth it for coding?

For coding, the data is mixed. Approval is 52% across 42 opinions, and 34 come from firsthand use. Code quality gets 14 positive and 6 negative mentions, while reliability gets 0 positive and 9 negative.

Is Gemini 3 Pro better than Claude Opus 4.5?

Among developers who named either model, the data favors Claude Opus 4.5. Four preferred Opus 4.5, and two preferred Gemini 3 Pro. The sample is small.

Is Gemini 3 Pro reliable?

Reliability is its weakest area in the data. It has 0 positive and 9 negative mentions. Developers report overload errors, rate limit errors, and occasional hangs.