Gemini 3.1 Pro Preview vs Kimi K2.5

Kimi K2.5 leads Gemini 3.1 Pro Preview on coding, 38.6 to 32.3 on the Vibe Coding Index, and Kimi K2.5 costs 4.2x less per output token ($2.85 vs $12.00 per million). Updated October 21, 2026; 93 benchmarks have a result for both.

Replace Gemini 3.1 Pro Preview Replace Kimi K2.5

Gemini 3.1 Pro PreviewGoogle

Coding: 32.3
Overall: 54.0
Output $/M: $12.00

Kimi K2.5Moonshot AI

Coding: 38.6
Overall: 44.8
Output $/M: $2.85

vs boards won of 93 shared

Gemini 3.1 Pro Preview or Kimi K2.5: which to pick

Pick Kimi K2.5 for coding: it scores 38.6 on the Vibe Coding Index against 32.3 for Gemini 3.1 Pro Preview, though on 3 benchmarks against 4.

Pick Gemini 3.1 Pro Preview for a larger context window (1.05M against 262K tokens) and work measured by LMArena WebDev or AA Coding Index, where it leads.

Model AGemini 3.1 Pro PreviewGoogle32.3/1004 independent benchmarks$0.99 per solved task Model BKimi K2.5Moonshot AI38.6/1003 independent benchmarks$0.24 per solved task

93406

Shared results: 93
Gemini 3.1 Pro Preview only: 40
Kimi K2.5 only: 6
Like-for-like categories: 8 of 9

Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.

Which one for your work

Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.

Coding
Vibe Coding Index, six public coding benchmarks
Kimi K2.5
Kimi K2.5 scores 38.6 against 32.3.
4 and 3 benchmarks behind the scores
Intelligence
AA Intelligence Index
Gemini 3.1 Pro Preview
Gemini 3.1 Pro Preview leads, 29.7 against 23.5.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Gemini 3.1 Pro Preview has a AA Agentic Index result, 8.2.
Directional only
Reasoning
GPQA Diamond
Gemini 3.1 Pro Preview
Gemini 3.1 Pro Preview leads, 95.5% against 84.1%.
Same board, both measured
Math
AIME
Gemini 3.1 Pro Preview
Gemini 3.1 Pro Preview leads, 98.1% against 95.6%.
Same board, both measured
Knowledge
MMLU Pro
Gemini 3.1 Pro Preview
Gemini 3.1 Pro Preview leads, 91.0% against 85.9%.
Same board, both measured
Web design
Design Arena: websites
Kimi K2.5
Kimi K2.5 leads, 1248 against 1234.
Same board, both measured
Long documents
Listed context window
Gemini 3.1 Pro Preview
Gemini 3.1 Pro Preview takes 1.05M tokens against 262K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Kimi K2.5
Gemini 3.1 Pro Preview $0.0080, Kimi K2.5 $0.0020 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
Kimi K2.5
Gemini 3.1 Pro Preview $0.99, Kimi K2.5 $0.24.
Published attempt cost and pass rate

At a glance

Coding index: Gemini 3.1 Pro Preview 32.3 from 4 benchmarks, Kimi K2.5 38.6 from 3 benchmarks.
Widest category: Reasoning, 15.7 points apart on average over 12 shared benchmarks.
Widest single gap: ARC-AGI-2: Gemini 3.1 Pro Preview 77.1% against Kimi K2.5 11.8%, 65.3 points apart on a 0 to 100 scale.
Output price: Gemini 3.1 Pro Preview $12.00 against Kimi K2.5 $2.85 per million output tokens, 4.2x apart.
Context window: Gemini 3.1 Pro Preview takes 1.05M tokens, Kimi K2.5 262K.

Benchmark by benchmark

First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.

Boards won, where both have a result
Category	Gemini 3.1 Pro Preview	Kimi K2.5
Coding	16	0
Intelligence	1	0
Agentic work	4	6
Reasoning	12	0
Math	3	0
Knowledge	10	2
Web design	3	5
General	29	2
All shared boards	78	15

CodingLike-for-likeSix of these boards feed the Vibe Coding Index160

Coding benchmarks for Gemini 3.1 Pro Preview and Kimi K2.5
Benchmark	Gemini 3.1 Pro Preview	Kimi K2.5	Leader
SWE-bench Verified	Not measured	70.8%	Only Kimi K2.5 measured
SWE-bench Multilingual	Not measured	67.3%	Only Kimi K2.5 measured
LMArena WebDev	1447	1437	Gemini 3.1 Pro Preview
AA Coding Index	68.8	46.8	Gemini 3.1 Pro Preview
LiveCodeBench (Vals AI)	88.5%	83.9%	Gemini 3.1 Pro Preview
SWE-bench Verified (Vals AI)	78.8%	70.0%	Gemini 3.1 Pro Preview
Terminal-Bench 4 (Vals AI)	2.5%	Not measured	Only Gemini 3.1 Pro Preview measured
Vibe Code Bench (Vals AI)	32.0%	17.5%	Gemini 3.1 Pro Preview
Code migration (Vals AI)	17.3%	7.0%	Gemini 3.1 Pro Preview
ProgramBench (Vals AI)	0.0%	Not measured	Only Gemini 3.1 Pro Preview measured
IOI (Vals AI)	51.8%	Not measured	Only Gemini 3.1 Pro Preview measured
Terminal-Bench 2.1 (Vals AI)	70.8%	42.0%	Gemini 3.1 Pro Preview
Vibe Code Bench 1-100 (Vals AI)	6.7%	Not measured	Only Gemini 3.1 Pro Preview measured
SciCode	58.7%	Not measured	Only Gemini 3.1 Pro Preview measured
Terminal-Bench Hard (Artificial Analysis run)	53.8%	34.8%	Gemini 3.1 Pro Preview
Terminal-Bench	80.2%	43.2%	Gemini 3.1 Pro Preview
GSO-Bench	22.6%	Not measured	Only Gemini 3.1 Pro Preview measured
WeirdML	72.1%	45.6%	Gemini 3.1 Pro Preview
MirrorCode	8.9%	Not measured	Only Gemini 3.1 Pro Preview measured
ALE-Bench	1160.6	821.6	Gemini 3.1 Pro Preview
LMArena Text: coding	1484	1483	Gemini 3.1 Pro Preview
LMArena Text: software and IT	1487	1471	Gemini 3.1 Pro Preview
LMArena WebDev: React	1438	1429	Gemini 3.1 Pro Preview
LMArena WebDev: HTML	1490	1426	Gemini 3.1 Pro Preview
LMArena WebDev: from an image	1481	1439	Gemini 3.1 Pro Preview
SciCode	14.3%	Not measured	Only Gemini 3.1 Pro Preview measured
Neither model has a result on the other 13 boards here

IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10

Intelligence benchmarks for Gemini 3.1 Pro Preview and Kimi K2.5
Benchmark	Gemini 3.1 Pro Preview	Kimi K2.5	Leader
AA Intelligence Index	29.7	23.5	Gemini 3.1 Pro Preview
Epoch Capabilities Index	Not measured	148.1	Only Kimi K2.5 measured

Agentic workLike-for-likePublic boards, shown as published, kept out of the index46

Agentic work benchmarks for Gemini 3.1 Pro Preview and Kimi K2.5
Benchmark	Gemini 3.1 Pro Preview	Kimi K2.5	Leader
AA Agentic Index	8.2	Not measured	Only Gemini 3.1 Pro Preview measured
Design Arena: agents building web apps	1125	1136	Kimi K2.5
Design Arena: agents building full-stack apps	1068	1114	Kimi K2.5
Design Arena: agents building mobile apps	1093	1147	Kimi K2.5
Design Arena: agents building Android apps	1048	1118	Kimi K2.5
Design Arena: agents building Godot games	1236	1219	Gemini 3.1 Pro Preview
Design Arena: agents building games	1114	Not measured	Only Gemini 3.1 Pro Preview measured
Design Arena: HTML slides	1153	Not measured	Only Gemini 3.1 Pro Preview measured
Design Arena: PowerPoint slides	1066	Not measured	Only Gemini 3.1 Pro Preview measured
SREBench (Vals AI)	0.0%	Not measured	Only Gemini 3.1 Pro Preview measured
Finance Agent (Vals AI)	43.0%	35.8%	Gemini 3.1 Pro Preview
Tax Agent Bench (Vals AI)	9.3%	Not measured	Only Gemini 3.1 Pro Preview measured
Legal Research Bench (Vals AI)	20.7%	15.9%	Gemini 3.1 Pro Preview
Excel Modeling Benchmark (Vals AI)	52.6%	28.5%	Gemini 3.1 Pro Preview
BioMysteryBench (Vals AI)	60.0%	Not measured	Only Gemini 3.1 Pro Preview measured
Terminal-Bench-Science (Vals AI)	1.4%	Not measured	Only Gemini 3.1 Pro Preview measured
Vals Science Index	27.4%	Not measured	Only Gemini 3.1 Pro Preview measured
τ²-Bench Telecom	95.6%	95.9%	Kimi K2.5
APEX-Agents	35.1%	Not measured	Only Gemini 3.1 Pro Preview measured
METR time horizons	77.0%	Not measured	Only Gemini 3.1 Pro Preview measured
BALROG	57.0%	Not measured	Only Gemini 3.1 Pro Preview measured
DeepResearch Bench	47.8%	Not measured	Only Gemini 3.1 Pro Preview measured
PostTrainBench	22.0%	Not measured	Only Gemini 3.1 Pro Preview measured
OSWorld	Not measured	63.3%	Only Kimi K2.5 measured
Vending-Bench 2	$911	$1,198	Kimi K2.5
Neither model has a result on the other 16 boards here

ReasoningLike-for-likePublic boards, shown as published, kept out of the index120

Reasoning benchmarks for Gemini 3.1 Pro Preview and Kimi K2.5
Benchmark	Gemini 3.1 Pro Preview	Kimi K2.5	Leader
GPQA Diamond	95.5%	84.1%	Gemini 3.1 Pro Preview
MysteryMechanism (Vals AI)	20.7%	Not measured	Only Gemini 3.1 Pro Preview measured
MMMU Pro (Vals AI)	88.2%	84.3%	Gemini 3.1 Pro Preview
GPQA Diamond (Artificial Analysis run)	94.1%	87.9%	Gemini 3.1 Pro Preview
Humanity's Last Exam (Artificial Analysis run)	47.0%	30.7%	Gemini 3.1 Pro Preview
GPQA Diamond (Epoch AI run)	94.4%	Not measured	Only Gemini 3.1 Pro Preview measured
ARC-AGI	98.0%	65.3%	Gemini 3.1 Pro Preview
ARC-AGI-2	77.1%	11.8%	Gemini 3.1 Pro Preview
SimpleBench	79.6%	46.8%	Gemini 3.1 Pro Preview
Chess puzzles	51.0%	Not measured	Only Gemini 3.1 Pro Preview measured
Mystery game puzzles	34.0%	Not measured	Only Gemini 3.1 Pro Preview measured
EBR-bench	14.3%	Not measured	Only Gemini 3.1 Pro Preview measured
LMArena Text: hard prompts	1485	1453	Gemini 3.1 Pro Preview
LMArena Text: expert	1483	1466	Gemini 3.1 Pro Preview
LMArena Text: hard prompts in English	1480	1460	Gemini 3.1 Pro Preview
LMArena Vision: diagrams	1311	1282	Gemini 3.1 Pro Preview
LMArena Vision: homework	1316	1297	Gemini 3.1 Pro Preview
Kagi LLM Benchmark	Not measured	78.5%	Only Kimi K2.5 measured
Neither model has a result on the other 4 boards here

MathLike-for-likePublic boards, shown as published, kept out of the index30

Math benchmarks for Gemini 3.1 Pro Preview and Kimi K2.5
Benchmark	Gemini 3.1 Pro Preview	Kimi K2.5	Leader
AIME	98.1%	95.6%	Gemini 3.1 Pro Preview
OTIS Mock AIME	95.6%	Not measured	Only Gemini 3.1 Pro Preview measured
FrontierMath, tiers 1 to 3	59.6%	Not measured	Only Gemini 3.1 Pro Preview measured
FrontierMath, tier 4	26.8%	Not measured	Only Gemini 3.1 Pro Preview measured
ProofBench	26.0%	Not measured	Only Gemini 3.1 Pro Preview measured
LMArena Text: math	1486	1470	Gemini 3.1 Pro Preview
LMArena Text: mathematical work	1478	1473	Gemini 3.1 Pro Preview
Neither model has a result on the other 6 boards here

KnowledgeLike-for-likePublic boards, shown as published, kept out of the index102

Knowledge benchmarks for Gemini 3.1 Pro Preview and Kimi K2.5
Benchmark	Gemini 3.1 Pro Preview	Kimi K2.5	Leader
MMLU Pro	91.0%	85.9%	Gemini 3.1 Pro Preview
LegalBench (Vals AI)	87.4%	Not measured	Only Gemini 3.1 Pro Preview measured
CaseLaw (Vals AI)	64.8%	58.7%	Gemini 3.1 Pro Preview
CorpFin (Vals AI)	64.5%	68.3%	Kimi K2.5
TaxEval (Vals AI)	72.9%	74.2%	Kimi K2.5
MortgageTax (Vals AI)	69.4%	66.5%	Gemini 3.1 Pro Preview
MedQA (Vals AI)	96.4%	94.4%	Gemini 3.1 Pro Preview
MedCode (Vals AI)	59.1%	39.3%	Gemini 3.1 Pro Preview
SimpleQA Verified	73.5%	34.3%	Gemini 3.1 Pro Preview
LMArena Text: science	1497	1459	Gemini 3.1 Pro Preview
LMArena Text: medicine and healthcare	1486	1448	Gemini 3.1 Pro Preview
LMArena Text: legal and government	1487	1441	Gemini 3.1 Pro Preview
LMArena Text: business and finance	1459	1427	Gemini 3.1 Pro Preview
FACTS Benchmark Suite	67.4%	Not measured	Only Gemini 3.1 Pro Preview measured
Neither model has a result on the other 4 boards here

Web designLike-for-likePublic boards, shown as published, kept out of the index35

Web design benchmarks for Gemini 3.1 Pro Preview and Kimi K2.5
Benchmark	Gemini 3.1 Pro Preview	Kimi K2.5	Leader
Design Arena: websites	1234	1248	Kimi K2.5
Design Arena: UI components	1227	1236	Kimi K2.5
Design Arena: game dev	1203	1207	Kimi K2.5
Design Arena: data viz	1209	1219	Kimi K2.5
Design Arena: 3D	1234	1205	Gemini 3.1 Pro Preview
Design Arena: all code categories	1231	1236	Kimi K2.5
Design Arena: SVG	1285	1149	Gemini 3.1 Pro Preview
Design Arena: ASCII art	1270	1166	Gemini 3.1 Pro Preview
Neither model has a result on the other 1 board here

GeneralLike-for-likePublic boards, shown as published, kept out of the index292

General benchmarks for Gemini 3.1 Pro Preview and Kimi K2.5
Benchmark	Gemini 3.1 Pro Preview	Kimi K2.5	Leader
Public Benefits Bench (Vals AI)	53.8%	Not measured	Only Gemini 3.1 Pro Preview measured
SAGE (Vals AI)	48.7%	49.9%	Kimi K2.5
MedScribe (Vals AI)	76.1%	76.4%	Kimi K2.5
Vals Multimodal Index	56.1%	Not measured	Only Gemini 3.1 Pro Preview measured
IFBench	77.1%	70.2%	Gemini 3.1 Pro Preview
Long context reasoning (AA-LCR)	82.0%	78.0%	Gemini 3.1 Pro Preview
DTBench	97.1%	Not measured	Only Gemini 3.1 Pro Preview measured
LMCA	53.8%	Not measured	Only Gemini 3.1 Pro Preview measured
Fiction.LiveBench	Not measured	86.1%	Only Kimi K2.5 measured
CL-bench	20.8%	19.3%	Gemini 3.1 Pro Preview
Furniture assembly	26.7%	Not measured	Only Gemini 3.1 Pro Preview measured
ExploitBench	26.1%	Not measured	Only Gemini 3.1 Pro Preview measured
CL-bench Life	16.9%	13.2%	Gemini 3.1 Pro Preview
LMArena Text	1480	1445	Gemini 3.1 Pro Preview
LMArena Text: instruction following	1466	1431	Gemini 3.1 Pro Preview
LMArena Text: creative writing	1482	1422	Gemini 3.1 Pro Preview
LMArena Vision	1296	1269	Gemini 3.1 Pro Preview
LMArena Text: English	1478	1453	Gemini 3.1 Pro Preview
LMArena Text: non-English	1476	1433	Gemini 3.1 Pro Preview
LMArena Text: Chinese	1527	1494	Gemini 3.1 Pro Preview
LMArena Text: German	1490	1441	Gemini 3.1 Pro Preview
LMArena Text: Japanese	1491	1423	Gemini 3.1 Pro Preview
LMArena Text: Korean	1455	1412	Gemini 3.1 Pro Preview
LMArena Text: French	1487	1454	Gemini 3.1 Pro Preview
LMArena Text: Spanish	1481	1450	Gemini 3.1 Pro Preview
LMArena Text: Russian	1497	1435	Gemini 3.1 Pro Preview
LMArena Text: Polish	1498	1445	Gemini 3.1 Pro Preview
LMArena Text: multi-turn	1487	1444	Gemini 3.1 Pro Preview
LMArena Text: longer queries	1482	1445	Gemini 3.1 Pro Preview
LMArena Text: writing and language	1481	1425	Gemini 3.1 Pro Preview
LMArena Text: entertainment and media	1465	1417	Gemini 3.1 Pro Preview
LMArena Vision: English	1287	1274	Gemini 3.1 Pro Preview
LMArena Vision: Chinese	1364	1328	Gemini 3.1 Pro Preview
LMArena Vision: reading text	1304	1276	Gemini 3.1 Pro Preview
LMArena Vision: humor	1322	1284	Gemini 3.1 Pro Preview
LMArena Vision: captioning	1345	1257	Gemini 3.1 Pro Preview
LMArena Vision: recognizing things	1330	1285	Gemini 3.1 Pro Preview
LMArena Vision: creative writing	1324	1269	Gemini 3.1 Pro Preview
LMArena Document	1444	Not measured	Only Gemini 3.1 Pro Preview measured
Neither model has a result on the other 7 boards here

Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared

Image generation benchmarks for Gemini 3.1 Pro Preview and Kimi K2.5
Benchmark	Gemini 3.1 Pro Preview	Kimi K2.5	Leader
Neither model has a result on 5 boards here

What each workload costs

Three fixed token mixes at each model's listed price, so per-token rates become a bill.

Chat turn

1K input + 500 output tokens

Gemini 3.1 Pro Preview$0.0080Fits in one request

Kimi K2.5$0.0020Fits in one request

Kimi K2.5 costs less

Repository review

50K input + 3K output tokens

Gemini 3.1 Pro Preview$0.136Fits in one request

Kimi K2.5$0.035Fits in one request

Kimi K2.5 costs less

Agent loop

200K input + 10K output tokens

Gemini 3.1 Pro Preview$0.520Fits in one request

Kimi K2.5$0.135Fits in one request

Kimi K2.5 costs less

Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.

Specification differences

The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.

Specifications of Gemini 3.1 Pro Preview and Kimi K2.5
Specification	Google Gemini 3.1 Pro Preview	MoonshotAI Kimi K2.5
Maker	Google	Moonshot AI
Weights	No public weights listed	Open weights
Released	February 19, 2026	January 27, 2026
Context window	1.05M	262K
Input $/Mtok	$2.00	$0.532
Output $/Mtok	$12.00	$2.85
Coding index	32.3	38.6
Independent measurements behind it	4	3
Cost per solved task	$0.99	$0.24
Source	OpenRouter listing	OpenRouter listing

Questions

Is Gemini 3.1 Pro Preview better than Kimi K2.5 for coding?

Pick Kimi K2.5 for coding: it scores 38.6 on the Vibe Coding Index against 32.3 for Gemini 3.1 Pro Preview, though on 3 benchmarks against 4. Pick Gemini 3.1 Pro Preview for a larger context window (1.05M against 262K tokens) and work measured by LMArena WebDev or AA Coding Index, where it leads.

Which is cheaper, Gemini 3.1 Pro Preview or Kimi K2.5?

Gemini 3.1 Pro Preview costs $2.00 per million input tokens and $12.00 per million output tokens; Kimi K2.5 costs $0.532 and $2.85. Per solved SWE-bench Verified task, Gemini 3.1 Pro Preview costs $0.99 and Kimi K2.5 $0.24.

Which has the bigger context window, Gemini 3.1 Pro Preview or Kimi K2.5?

Gemini 3.1 Pro Preview, with 1.05M tokens against 262K.

How many benchmarks is the Gemini 3.1 Pro Preview vs Kimi K2.5 comparison based on?

93 benchmarks have a result for both models as of October 21, 2026. Gemini 3.1 Pro Preview's coding index rests on 4 independent benchmarks and Kimi K2.5's on 3. Where only one model has a result, the page says so instead of naming a winner.

Compare next

Embed this comparison on your site

Free under CC BY 4.0, while the source link stays visible. Preview the frame · Terms and other embeds

<iframe src="https://llmbenchmark.ai/embed/compare/gemini-3-1-pro-preview-vs-kimi-k2-5" title="Gemini 3.1 Pro Preview vs Kimi K2.5, LLMBenchmark.ai" width="100%" height="360" style="border:0;max-width:100%" loading="lazy"></iframe>
<p style="font:13px/1.4 sans-serif;margin:6px 0 0">Source: <a href="https://llmbenchmark.ai/compare/gemini-3-1-pro-preview-vs-kimi-k2-5">LLMBenchmark.ai, Gemini 3.1 Pro Preview vs Kimi K2.5</a></p>

Gemini 3.1 Pro Preview or Kimi K2.5: which to pick

Which one for your work

Coding

Intelligence

Agentic work

Reasoning

Math

Knowledge

Web design

Long documents

Everyday cost

Cost per solved task

At a glance

Benchmark by benchmark

What each workload costs

Chat turn

Repository review

Agent loop

Specification differences

Questions

Compare next