IonWarpStarter
Try AI PR review.
- 2 reviewers
- Code Review
- Security Review
- 3 seats included
- 15,000 credits to start, then 5,000/mo
- Unlimited repos
Gemini code review
Gemini code review: Gemini Code Assist on GitHub, and Gemini 3.8 Flash on our benchmarks
Google's Gemini Code Assist reviews GitHub pull requests: it summarizes each one, posts an in-depth review, and answers follow-up questions when you tag /gemini in a comment. Google documents a version set up through Google Cloud for GitHub and GitHub Enterprise, in Preview, with a quota of 100+ pull requests a day.[1]
We measured the model itself. Gemini 3.8 Flash reviewed the same public pull requests as DeepSeek, GLM, Muse Spark and GPT models and four review apps, and a model grader matched each finding to a known issue. On docs review it scored an F1 of 68.6%, the best of every model and app we ran.[2]
Where Gemini 3.8 Flash leads, and where it does not
Docs review: first of every reviewer
12 of 17 known documentation gaps found on 19 pull requests, with 6 findings that matched none, for about 8 cents a review in a median minute. DeepSeek V4 Pro scored 65.2% and the best app, CodeRabbit, 55.6%.[2]
Docs ReviewOn Pro and Max
Logs and errors: the top model
67.7% on 17 pull requests with 31 known issues: 21 found, 10 that matched none, about 10 cents a review. Two apps scored higher, Qodo (72.7%) and Greptile (67.8%); DeepSeek V4.1 Flash scored 59.0% at 4 cents.[3]
Log & ErrorsOn Pro and Max
Security: mid-table, and the costliest
51.6% on 13 pull requests, 8 of 18 known issues found, at about 52 cents a review and a median of 7 minutes, against 66.7% for GLM 5.3 Flash at about a cent.[4]
API performance and AI architecture: behind
41.7% on 18 API performance pull requests, last of the ten reviewers we ran,[5] and 33.3% on the 11-pull-request AI architecture set.[6]
How Gemini scored on each review type
| Review | Model | F1 | Precision | Recall | $ per review | Median time | Rank |
|---|---|---|---|---|---|---|---|
| Security Review13 of 13 PRs | Gemini 3.8 Flash | 51.6% | 61.5% | 44.4% | $0.52 | 7m 10s | 4 of 7 |
| AI Architecture Review11 of 11 PRs | Gemini 3.8 Flash | 33.3% | 25.0% | 50.0% | $0.24 | 3m 50s | 6 of 8 |
| Log & Errors Review17 of 17 PRs | Gemini 3.8 Flash | 67.7% | 67.7% | 67.7% | $0.10 | 1m 28s | 1 of 5 |
| Docs Review19 of 19 PRs | Gemini 3.8 Flash | 68.6% | 66.7% | 70.6% | $0.08 | 59 sec | 1 of 3 |
| API Performance Review18 of 18 PRs | Gemini 3.8 Flash | 41.7% | 45.5% | 38.5% | $0.07 | 47 sec | 6 of 6 |
Rank is among the models we ran on that review, by F1; a tie goes to the cheaper one. See the benchmark
Simple pricing for your whole team.
One workspace price with seats included. Credits pool across your team. Unlimited repos on every plan.
Pro
Most popularGet AI PR review right from day 1.
- +5 reviewers
- DevOps Review
- Docs Review
- AI Architecture
- Log & Errors
- Cost Review
- 5 seats included
- +25,000 credits/mo
- +$29/mo per added seat
- Unlimited repos
Max
Scale AI PR review.
- +9 reviewers
- UX Review
- API Performance
- Analytics Review
- SEO Review
- Changelog
- CI/CD Performance
- Plan Review
- Repeat-Failure Learnings
- SOC 2 Review
- 10 seats included
- +75,000 credits/mo
- +$29/mo per added seat
- Unlimited repos
- Priority support
Enterprise
Custom reviewers + credits · GitHub Enterprise · SSO/SAML + audit logs · Custom security review + SLA
Frequently asked questions
Is Gemini good for code review?
For some reviews. On our benchmarks Gemini 3.8 Flash was the best reviewer we ran on docs (68.6%) and the best model on logs and errors (67.7%), but it trailed on API performance and AI architecture, and its security reviews cost about 52 cents each. We have not run it on general code review.
How do I get Gemini to review a GitHub pull request?
Set up Gemini Code Assist from the Google Cloud console, with a project that has billing, and install its app on your GitHub organization or account. It then summarizes and reviews pull requests, and answers follow-ups when you tag /gemini in a comment.
Can IonWarp review with Gemini?
Yes. Docs Review and Log & Errors retry on Gemini 3.8 Flash, the best-scoring model on both of their benchmarks, and a workspace can run it on those reviewers from the Reviewers page. Their default, DeepSeek V4.1 Flash, costs less per review.
What does a Gemini code review cost?
In model fees on our benchmarks, Gemini 3.8 Flash cost about 7 cents a review on API performance, 8 on docs, 10 on logs and errors, and 52 on security. IonWarp plans include the model cost: Pro is $49/mo for the workspace.
Which Gemini models has IonWarp benchmarked?
Gemini 3.8 Flash, on five review types: docs, logs and errors, security, API performance and AI architecture. It is not in our general code review benchmark.
Sources
- Review GitHub code using Gemini Code Assist
- The docs review benchmark: 19 pull requests, 17 known issues (release of 2026-09-22)
- The log and error review benchmark: 17 pull requests, 31 known issues (release of 2026-09-24)
- The security review benchmark: 13 pull requests, 18 known issues (release of 2026-09-24)
- The API performance review benchmark: 18 pull requests, 13 known issues (release of 2026-09-24)
- The AI architecture review benchmark: 11 pull requests, 6 known issues (release of 2026-09-24)