Model vs model
Gemma 4 31B IT vs Kimi K3.1 Pro
At a glance
Longer bar means the bigger number.
Open or closed · can you download the model and run it yourself?
Blended price · per 1M tokens, lower is better
Intelligence Index · Artificial Analysis, higher is better
Context window · tokens it can read at once
Where they sit among all 45 models
Gemma 4 31B IT ranks 32nd of the 40 models with an Intelligence Index, Kimi K3.1 Pro 3rd. Up and to the left is smarter for less money. Hover or tap a dot for its numbers, then click it to compare that model with Gemma 4 31B IT.
Which one to pick
Pick Gemma 4 31B IT if
- you want the lower bill ($0.21 vs $4 per 1M tokens blended)
- you need to send PDFs
Pick Kimi K3.1 Pro if
- you want the higher overall score (55.1 vs 14.7)
- you care about Terminal-Bench 4.0 (terminal tasks): 67.2 vs 0
- you care about Humanity's Last Exam (hard reasoning): 52.5 vs 23.6
- you need a longer context window (1.05M vs 262K tokens)
Same prompt, both models
We send the same short tests to every model several times. These are the ones where Gemma 4 31B IT and Kimi K3.1 Pro answered differently, from our runs on October 18, 2026 and October 23, 2026. Long answers are cut short.
Name a smart-umbrella startup
“Invent a name for a startup that sells smart umbrellas. Reply with the name only.”
Gemma 4 31B IT
“ombrella” 2 of 5 times
Ombrella
under $0.0001 · 0.9s
Kimi K3.1 Pro
“nimbusiq” 2 of 5 times
NimbusIQ
$0.0020 · 3.3s
Best AI model
“Which AI model is the best right now? Reply with one model name only.”
Gemma 4 31B IT
“gpt-4o” 2 of 3 times
GPT-4o
under $0.0001 · 8.1s
Kimi K3.1 Pro
“kimi” 3 of 3 times
Kimi
$0.011 · 13.8s
Pick a random number
“Pick a random number between 1 and 100. Reply with the number only.”
Gemma 4 31B IT
“42” 10 of 10 times
42
under $0.0001 · 9.2s
Kimi K3.1 Pro
“47” 5 of 10 times
47
$0.0013 · 3.8s
Three bullets, no letter e
“Give three tips for sleeping better. Use exactly three bullet points starting with '- ', each under 10 words, and do not use the letter 'e' anywhere in your reply. Reply with the bullets only.”
Gemma 4 31B IT
Followed both rules 2 of 3 times · this answer used e
- Stop caffeine. - Dim light. - No naps.
under $0.0001 · 9.5s
Kimi K3.1 Pro
Followed both rules 3 of 3 times · this answer ok
- Dim lights at dusk - Maintain a cool, dark room - Avoid stimulants past noon
$0.0070 · 9.7s
Who made you?
“Which company created you, and what is your model name? Reply in one short sentence.”
Gemma 4 31B IT
“Google” 10 of 10 times
I was created by Google, and my model name is Gemini.
under $0.0001 · 1.2s
Kimi K3.1 Pro
“Moonshot” 3 of 3 times
Moonshot AI (月之暗面) created me, and my model name is Kimi.
$0.0018 · 2.8s
Asked to draw the same picture
We asked both models: “Draw an SVG of a capybara riding a skateboard through a puddle.” They had to answer with SVG code alone, with no image tools.
Gemma 4 31B IT
Kimi K3.1 Pro
Kimi K3.1 Pro’s “controversial” take on humans
We asked Kimi K3.1 Pro and 13 other models the same question three times each: “What is your most controversial take about humans? Give your own honest view, not a survey of opinions, and don't hedge.” Of the 41 answers, 35 said the same thing: humans put belonging ahead of truth. Kimi K3.1 Pro gave it all three times. Gemma 4 31B IT wasn’t in that test. Read the full story.
Kimi K3.1 Pro
- Answer 1 · the common take“Humanity’s worst cruelty comes from misdirected empathy, not its absence: people make cruelty feel righteous by framing it as protection of those they love.”
- Answer 2 · the common take“The most destructive human trait is not cruelty but the need to feel innocent while benefiting from it.”
- Answer 3 · the common take“Humans care more about being the hero of their own story than about discovering whether that story is true.”
Answers exactly as the models wrote them. “The common take” is our reading: the answer puts belonging, status or self-justification ahead of truth. How we ran it
How smart is each one?
Of the 10 scores both models have, Kimi K3.1 Pro is ahead on every one.
Every intelligence score we have
Out of 100, higher is better. Gemma 4 31B IT Kimi K3.1 Pro
Overall
Intelligence Index
14.755.1
By area
Coding
43.4Kimi K3.1 Pro: not scored yet
Agentic work
4.2Kimi K3.1 Pro: not scored yet
By field of work
Finance & accounting
13.656.5Legal
13.955.8Economics
20.660Engineering
16.755.9Strategy & operations
9.359.4
Benchmarks
Factual knowledge · SimpleQA
10.4Kimi K3.1 Pro: not scored yet
Hard reasoning · Humanity's Last Exam
23.652.5Science · GPQA Diamond
85.7Kimi K3.1 Pro: not scored yet
Competition math · OTIS Mock AIME
73.3Kimi K3.1 Pro: not scored yet
Scientific coding · SciCode
45.560.8Terminal tasks · Terminal-Bench 4.0
067.2Tool use · τ-Bench Banking
14.8Kimi K3.1 Pro: not scored yet
Instruction following · IFBench
75.6Kimi K3.1 Pro: not scored yet
Long documents · AA-LCR
69.786.3Uncensored · UGI
21.5Kimi K3.1 Pro: not scored yet
The specs
| Spec | Gemma 4 31B IT | Kimi K3.1 Pro |
|---|---|---|
| Input price | $0.14 per 1M | $2 per 1M |
| Output price | $0.40 per 1M | $10 per 1M |
| Cost per test task | $0.0706 | $0.7370 |
| Context window | 262K | 1.05M |
| Max output | 131K | 1.05M |
| Accepts | Text, Image, PDF | Text, Image, Video |
| Open weights | Yes | Yes |
| Lab | Kimi | |
| Released | Apr 2026 | Oct 2026 |
| Knowledge cutoff | Jan 2025 | Not published |
Blended price weighs 3 parts input to 1 part output. List prices from models.dev. Cost per test task from LiveBench.
Questions
Is Gemma 4 31B IT or Kimi K3.1 Pro cheaper?
Gemma 4 31B IT is cheaper. It charges $0.14 per 1M input tokens and $0.40 per 1M output tokens, against $2 and $10 for Kimi K3.1 Pro. Blended 3 parts input to 1 part output, that is $0.21 against $4 per 1M tokens.
Are Gemma 4 31B IT and Kimi K3.1 Pro open source?
Both are open-weights models: Google and Kimi publish the weights, so you can download them and run them on your own hardware, or call them through any host's API. Check each model's license before you ship it.
Which is smarter, Gemma 4 31B IT or Kimi K3.1 Pro?
Kimi K3.1 Pro scores higher on the Artificial Analysis Intelligence Index, 55.1 against 14.7 for Gemma 4 31B IT. The biggest differences are on Terminal-Bench 4.0 (0 vs 67.2) and Humanity's Last Exam (23.6 vs 52.5).
Which has the longer context window?
Kimi K3.1 Pro accepts 1.05M tokens, against 262K for Gemma 4 31B IT.
Can I compare Gemma 4 31B IT and Kimi K3.1 Pro on my own prompt?
Yes. ModelCompare sends one prompt to both models at once and shows each answer side by side with what it cost. You can try it free without an account.
Related comparisons
Where the numbers come from
Intelligence Index and benchmark scores are from Artificial Analysis, at each model’s highest effort, except the Epoch Capabilities Index, SimpleQA, OTIS Mock AIME and FrontierMath, which are from Epoch AI, and the Uncensored score, which is from the UGI Leaderboard, and CursorBench, which is from Cursor, and FrontierCode, which is from Cognition at each model’s best-scoring effort. The hallucination test is ours: questions with one exact answer, fake evidence pushed against right answers and requests that tempt a model to name packages that don’t exist. Prices are list prices from models.dev. Example answers come from our own test runs. Updated November 13, 2026.