Compare · ModelsLive · 2 picked · head to head
Gemini 3.8 Flash vs Kimi K3
Side by side · benchmarks, pricing, and signals you can act on.
Winner summary
Gemini 3.8 Flash wins on 15/27 benchmarks
Gemini 3.8 Flash wins 15 of 27 shared benchmarks. Leads in agentic · reasoning · knowledge.
Category leads
speed·Kimi K3agentic·Gemini 3.8 Flashreasoning·Gemini 3.8 Flashknowledge·Gemini 3.8 Flashcoding·Gemini 3.8 Flashgeneral·Gemini 3.8 Flashmath·Kimi K3
Hype vs Reality
Attention vs performance
Gemini 3.8 Flash
#53 by perf·#11 by attention
Kimi K3
#54 by perf·#16 by attention
Best value
Gemini 3.8 Flash
2.5x better value than Kimi K3
Gemini 3.8 Flash
24.1 pts/$
$2.48/M
Kimi K3
9.8 pts/$
$6.09/M
Vendor risk
Who is behind the model
Google DeepMind
$4.09T·Tier 1
Moonshot AI
$50.0B·Tier 1
Head to head
27 benchmarks · 2 models
Gemini 3.8 FlashKimi K3
Artificial Analysis · CritPt
Kimi K3 leads by +5.1
Gemini 3.8 Flash
18.3
Kimi K3
23.4
Artificial Analysis · GDPval
Kimi K3 leads by +4.9
Gemini 3.8 Flash
46.8
Kimi K3
51.7
Artificial Analysis · GPQA Diamond
Gemini 3.8 Flash leads by +1.8
Gemini 3.8 Flash
95.3
Kimi K3
93.5
Artificial Analysis · Humanity's Last Exam
Gemini 3.8 Flash leads by +0.9
Gemini 3.8 Flash
47.8
Kimi K3
46.9
Artificial Analysis · Long Context Reasoning
Kimi K3 leads by +7.4
Gemini 3.8 Flash
81.3
Kimi K3
88.7
Artificial Analysis · MMMU Pro
Gemini 3.8 Flash leads by +5.1
Gemini 3.8 Flash
85.6
Kimi K3
80.5
Artificial Analysis · Quality Index
Kimi K3 leads by +2.7
Gemini 3.8 Flash
40.9
Kimi K3
43.6
Artificial Analysis · SciCode
Kimi K3 leads by +2.9
Gemini 3.8 Flash
56.6
Kimi K3
59.5
APEX-Agents
Gemini 3.8 Flash leads by +13.7
APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments.
Gemini 3.8 Flash
64.3
Kimi K3
50.6
ARC-AGI
Gemini 3.8 Flash leads by +4.0
ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization.
Gemini 3.8 Flash
98.5
Kimi K3
94.5
ARC-AGI-2
Gemini 3.8 Flash leads by +28.8
ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data.
Gemini 3.8 Flash
89.2
Kimi K3
60.4
Chess Puzzles
Gemini 3.8 Flash leads by +22.1
Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities.
Gemini 3.8 Flash
55.8
Kimi K3
33.7
Deepswe
Gemini 3.8 Flash leads by +5.3
Gemini 3.8 Flash
73.8
Kimi K3
68.5
Dtbench
Gemini 3.8 Flash leads by +7.5
Gemini 3.8 Flash
92.9
Kimi K3
85.3
Frontiercode
Kimi K3 leads by +3.0
Gemini 3.8 Flash
41.2
Kimi K3
44.2
FrontierMath-Tier-4-v2-Private
Kimi K3 leads by +17.1
Gemini 3.8 Flash
21.9
Kimi K3
39.0
FrontierMath-Tiers-1-3-v2-Private
Kimi K3 leads by +3.8
Gemini 3.8 Flash
68.4
Kimi K3
72.2
Frontierswe
Kimi K3 leads by +6.2
Gemini 3.8 Flash
19.6
Kimi K3
25.9
Furniture Assembly
Kimi K3 leads by +3.6
Gemini 3.8 Flash
2.4
Kimi K3
6.0
GPQA diamond
Gemini 3.8 Flash leads by +3.0
Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs.
Gemini 3.8 Flash
93.9
Kimi K3
90.8
Lmca
Gemini 3.8 Flash leads by +0.2
Gemini 3.8 Flash
62.2
Kimi K3
62.0
Mystery Game Puzzles
Gemini 3.8 Flash leads by +23.1
Gemini 3.8 Flash
41.6
Kimi K3
18.5
OTIS Mock AIME 2024-2025
Gemini 3.8 Flash leads by +1.7
OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills.
Gemini 3.8 Flash
98.9
Kimi K3
97.2
Proofbench
Kimi K3 leads by +39.0
Gemini 3.8 Flash
48.0
Kimi K3
87.0
SimpleQA Verified
Gemini 3.8 Flash leads by +19.1
SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information.
Gemini 3.8 Flash
69.7
Kimi K3
50.6
Surface Evolver Bench
Kimi K3 leads by +18.1
Gemini 3.8 Flash
76.9
Kimi K3
95.0
WeirdML
Gemini 3.8 Flash leads by +2.2
WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns.
Gemini 3.8 Flash
84.8
Kimi K3
82.6
Full benchmark table
| Benchmark | Gemini 3.8 Flash | Kimi K3 |
|---|---|---|
Artificial Analysis · CritPt | 18.3 | 23.4 |
Artificial Analysis · GDPval | 46.8 | 51.7 |
Artificial Analysis · GPQA Diamond | 95.3 | 93.5 |
Artificial Analysis · Humanity's Last Exam | 47.8 | 46.9 |
Artificial Analysis · Long Context Reasoning | 81.3 | 88.7 |
Artificial Analysis · MMMU Pro | 85.6 | 80.5 |
Artificial Analysis · Quality Index | 40.9 | 43.6 |
Artificial Analysis · SciCode | 56.6 | 59.5 |
APEX-Agents APEX-Agents · evaluates AI agents on complex, multi-step tasks requiring planning, tool use, and autonomous decision-making in realistic environments. | 64.3 | 50.6 |
ARC-AGI ARC-AGI · the original Abstraction and Reasoning Corpus, testing whether AI can solve novel visual pattern recognition tasks without memorization. | 98.5 | 94.5 |
ARC-AGI-2 ARC-AGI-2 · the second iteration of the Abstraction and Reasoning Corpus, testing novel pattern recognition and abstract reasoning without prior training data. | 89.2 | 60.4 |
Chess Puzzles Chess Puzzles · tests strategic and tactical reasoning by having models solve chess puzzle positions, evaluating lookahead and pattern recognition abilities. | 55.8 | 33.7 |
Deepswe | 73.8 | 68.5 |
Dtbench | 92.9 | 85.3 |
Frontiercode | 41.2 | 44.2 |
FrontierMath-Tier-4-v2-Private | 21.9 | 39.0 |
FrontierMath-Tiers-1-3-v2-Private | 68.4 | 72.2 |
Frontierswe | 19.6 | 25.9 |
Furniture Assembly | 2.4 | 6.0 |
GPQA diamond Graduate-Level Google-Proof QA (Diamond set) · expert-crafted questions in physics, biology, and chemistry that are difficult even for domain PhDs. | 93.9 | 90.8 |
Lmca | 62.2 | 62.0 |
Mystery Game Puzzles | 41.6 | 18.5 |
OTIS Mock AIME 2024-2025 OTIS Mock AIME 2024-2025 · simulated American Invitational Mathematics Examination problems testing advanced problem-solving skills. | 98.9 | 97.2 |
Proofbench | 48.0 | 87.0 |
SimpleQA Verified SimpleQA Verified · short factual questions with verified answers, measuring factual accuracy and the tendency to hallucinate or provide incorrect information. | 69.7 | 50.6 |
Surface Evolver Bench | 76.9 | 95.0 |
WeirdML WeirdML · tests models on unusual and adversarial machine learning tasks that require creative problem-solving beyond standard patterns. | 84.8 | 82.6 |
Pricing · per 1M tokens · projected $/mo at 10M tokens
| Model | Input | Output | Context | Projected $/mo |
|---|---|---|---|---|
| $0.82 | $4.13 | 1.0M tokens (~524 books) | $16.50 | |
| $0.24 | $11.94 | 1.0M tokens (~524 books) | $31.63 |