Gemini 3.7 Flash vs Kimi K3
Full profiles: Gemini 3.7 Flash (Google) and Kimi K3 (Moonshot AI): every benchmark, price and rank.
Updated · Data from 11 sources
As of 25 Oct 2026, Gemini 3.7 Flash (high) and Kimi K3 (max) are neck and neck on quality (63.5 vs 64.8 on the BenchLeader Index, inside the ±4.8 margin of error), and Gemini 3.7 Flash (high) costs 4.0× less, which makes it the better buy for most workloads. Pick Gemini 3.7 Flash for knowledge (68 vs 63); pick Kimi K3 for agents & tools (59 vs 48).
Gemini 3.7 Flash benchmarks high Google
Kimi K3 benchmarks max Moonshot AIOpen in the comparison toolWhich to pick for…
The pick's numbers come first. Category scores on the BenchLeader scale (50 is the average model, 15 points one standard deviation). High confidence means a lead of half a standard deviation or more; under one point is called either.
| Use case | Pick | Why | Confidence |
|---|---|---|---|
| Coding | Kimi K3 | 65 vs 63 category score; Arena WebDev 1655 vs 1590 | Medium |
| Agents & tool use | Kimi K3 | 59 vs 48 category score; τ³-Banking (AA) 46.0% vs 32.8% | High |
| Maths | Either | 63 vs 63 category score; FrontierMath Tiers 1–3 71.6% vs 72.2% | Either |
| Reasoning | Either | 67 vs 66 category score; GPQA Diamond 94.8% vs 93.1% | Either |
| Knowledge | Gemini 3.7 Flash | 68 vs 63 category score; SimpleQA Verified 69.2% vs 50.6% | Medium |
| Long documents | Kimi K3 | 70 vs 66 category score; AA-LCR 88.7% vs 81.7% | Medium |
| Images and other media | Gemini 3.7 Flash | 68 vs 64 category score; MMMU-Pro 85.5% vs 80.5% | Medium |
| Chat quality (human preference) | Either | 68 vs 68 on Arena votes | Either |
| Lowest cost | Gemini 3.7 Flash | $1.50 vs $6.00 per 1M blended; agentic coding session $0.060 vs $0.240 | – |
| Fastest first answer | Gemini 3.7 Flash | 2.5 s (Gemini 3.7 Flash (high)) vs 52.6 s (Kimi K3 (max)) to the first answer token | – |
| High volume on a budget | Neither flagship | A smaller model is usually the better fit: Cheapest good LLMs, Fastest LLMs | – |
Key differences
- Gemini 3.7 Flash (high) and Kimi K3 (max) are neck and neck on quality (63.5 vs 64.8, within the ±4.8 margin).
- Gemini 3.7 Flash (high) is stronger in human preference, knowledge, multimodal, reasoning.
- Kimi K3 (max) is stronger in agents & tools, coding, composite, long context, maths.
- Gemini 3.7 Flash (high) is 4.0× cheaper ($1.50 vs $6.00 per 1M blended).
- Gemini 3.7 Flash (high) streams 2.2× faster (88 vs 41 tokens per second).
At a glance
| Spec | Gemini 3.7 Flash | Kimi K3 |
|---|---|---|
| Developer | Moonshot AI | |
| Released | 13 Aug 2026 | 16 Jul 2026 |
| Context window (tokens) | 1,048,576 | 1,048,576 |
| Compared at | Gemini 3.7 Flash (high) | Kimi K3 (max) |
| Open weights | No | Yes |
| Price in / out / cached input per 1M | $0.750 / $3.75 / $0.075 | $3.00 / $15.00 / $0.300 |
Category by category
Category scores on the BenchLeader scale: 50 is the average evaluated model, 15 points is one standard deviation. A gap under one point is called level.
| Category | Gemini 3.7 Flash (high) | Kimi K3 (max) | Leader |
|---|---|---|---|
| Reasoning | 67.3 | 66.4 | level |
| Coding | 62.9 | 64.6 | Kimi K3 |
| Agents & tools | 47.6 | 59.4 | Kimi K3 |
| Maths | 63.3 | 63.4 | level |
| Knowledge | 68.0 | 62.8 | Gemini 3.7 Flash |
| Human preference | 68.4 | 68.1 | level |
| Multimodal | 67.6 | 63.6 | Gemini 3.7 Flash |
| Long context | 65.9 | 69.6 | Kimi K3 |
| Composite | 73.2 | 78.3 | Kimi K3 |
Where sources disagree
Across the 59 benchmarks both have, Gemini 3.7 Flash leads 30, Kimi K3 leads 28 and 1 is a tie. Independent sources split: Gemini 3.7 Flash leads on Arena Text; Kimi K3 leads on AA Intelligence Index v4.3.2.
| Overall index | Gemini 3.7 Flash (high) | Kimi K3 (max) | Leader |
|---|---|---|---|
| BenchLeader Index | 63.5 | 64.8 | neck and neck |
| AA Intelligence Index v4.3.2 | 39.1 | 43.6 | Kimi K3 |
| Arena Text | 1490 | 1487 | Gemini 3.7 Flash |
Each publisher runs its own tasks, harnesses and effort settings, so their overall verdicts can differ; BenchLeader reconciles them on one scale. † measured at another effort setting of the same model (the publisher did not test the one compared here).
Price, speed and latency
List prices blended at three input tokens per output token. Per-task prices use the token counts on the methodology page; suite cost is what Artificial Analysis paid to run its whole Intelligence Index.
| Measure | Gemini 3.7 Flash (high) | Kimi K3 (max) |
|---|---|---|
| Blended price per 1M tokens | $1.50 | $6.00 |
| Input / output per 1M | $0.750 / $3.75 | $3.00 / $15.00 |
| Cached input per 1M | $0.075 | $0.300 |
| Price per task: chat reply | $0.0014 | $0.0057 |
| Price per task: summarize a 30-page report | $0.011 | $0.045 |
| Price per task: code edit | $0.010 | $0.041 |
| Price per task: agentic coding session | $0.060 | $0.240 |
| Price per task: structured extraction | $0.0023 | $0.0090 |
| Cost to run a full benchmark suite | – | $2.00 |
| Output speed | 88 tok/s | 41 tok/s |
| Time to first answer | 2.5 s | 52.6 s |
| Time to a complete answer | – | 65.0 s |
Each model is compared at its best-ranked reasoning-effort setting. Gemini 3.7 Flash (high) scores 63.5 and takes 2 s to its first answer token; Gemini 3.7 Flash (medium) scores 63.1 in 2 s, Gemini 3.7 Flash (low) scores 60.3 in 2 s. Kimi K3 (max) scores 64.8 and takes 53 s to its first answer token; Kimi K3 scores 58.8 in 1 s, Kimi K3 (low) scores 58.2 in 57 s. Higher effort means more thinking before the answer: better scores, longer waits and more tokens billed. How reasoning effort changes scores and cost.
Every benchmark
137 benchmarks, grouped by category; each name links to the board and its leaders. † measured at another effort setting of the same model, shown for reference and not counted as a win.
| Metric | Publisher | Gemini 3.7 Flash (high) | Kimi K3 (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 63.5 | 64.8 |
| Agents & tools score | BenchLeader | 47.6 | 59.4 |
| Coding score | BenchLeader | 62.9 | 64.6 |
| Composite score | BenchLeader | 73.2 | 78.3 |
| Human preference score | BenchLeader | 68.4 | 68.1 |
| Knowledge score | BenchLeader | 68.0 | 62.8 |
| Long context score | BenchLeader | 65.9 | 69.6 |
| Maths score | BenchLeader | 63.3 | 63.4 |
| Multimodal score | BenchLeader | 67.6 | 63.6 |
| Reasoning score | BenchLeader | 67.3 | 66.4 |
| Blended price $/M | BenchLeader | $1.50 | $6.00 |
| Output speed | BenchLeader | 88 tok/s | 41 tok/s |
| Time to first answer | BenchLeader | 2.5 s | 52.6 s |
| Context window | BenchLeader | 1,048,576 tokens | 1,048,576 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 94.8% | 93.1% |
| SimpleBench | SimpleBench | – | 60.7% |
| Arena Hard Prompts | Arena (LMArena) | 1509 | 1516 |
| LiveBench 2.0 Reasoning | LiveBench | 87.8%† | 90.7%† |
| LiveBench 1.0 Reasoning | LiveBench | 87.8% | 90.7%† |
| GPQA Diamond (AA) | Artificial Analysis | 94.5% | 93.5% |
| Humanity's Last Exam (AA) | Artificial Analysis | 47.9% | 46.9% |
| GPQA Diamond (Vals) | Vals AI | 93.9% | 92.9%† |
| ARC-AGI-1 | ARC Prize | 95.5% | 94.5% |
| ARC-AGI-2 | ARC Prize | 84.6% | 60.4% |
| CritPt | Artificial Analysis | 14.3% | 23.4% |
| Chess Puzzles | Epoch AI Benchmarking Hub | 66.0% | 36.0% |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 37.0% | 26.0% |
| LMCA | Epoch AI Benchmarking Hub | 50.4% | 52.7% |
| DTBench | Epoch AI Benchmarking Hub | 96.8% | 91.2% |
| ForecastBench | ForecastBench | 66.2% | 64.6% |
| Coding | |||
| SciCode | SciCode | 57.2% | 59.5% |
| WeirdML | WeirdML | – | 82.6% |
| FrontierCode | Cognition | 43.6%† | 44.2%† |
| Arena Coding | Arena (LMArena) | 1520 | 1539 |
| Arena WebDev | Arena (LMArena) | 1590 | 1655 |
| LiveBench 2.0 Coding | LiveBench | 80.5%† | 59.3%† |
| LiveBench 1.0 Coding | LiveBench | 78.9% | 81.5%† |
| LiveCodeBench | Vals AI | 88.7% | 87.2% |
| IOI | Vals AI | 67.8% | 48.9% |
Show all 137 benchmarks
| Metric | Publisher | Gemini 3.7 Flash (high) | Kimi K3 (max) |
|---|---|---|---|
| Coding | |||
| SWE-bench (Vals) | Vals AI | 80.8% | 93.4%† |
| Code Migration | Vals AI | 34.8% | 16.1% |
| Code Migration v1.1 | Vals AI | – | 28.6% |
| ProgramBench | Vals AI | 0.0% | 2.0%† |
| SWE-bench Multimodal | Vals AI | – | 44.2% |
| Vibe Code Bench 1-100 | Vals AI | – | 18.2% |
| Vibe Code Bench v1.1 | Vals AI | 70.4% | 85.0% |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | – | 93.0% |
| DeepSWE v1.1 | Epoch AI Benchmarking Hub | 65.3% | 68.5% |
| ALE-Bench | Epoch AI Benchmarking Hub | 904.3 | 1524.5 |
| FrontierSWE | Epoch AI Benchmarking Hub | 20.3% | 25.9% |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 13.6% | 12.6% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | 85.8% | 85.0% |
| Startrise Frontend | Startrise | – | 79.8%† |
| FrontierCode 1.1 (AA) | Artificial Analysis | – | 43.6% |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 11.2% | – |
| Remote Labor Index | Scale AI / CAIS | 5.0%† | – |
| APEX-Agents | Mercor | 67.8%† | 47.8%† |
| Arena Agent | Arena (LMArena) | -1.3 | 1.6 |
| LiveBench Long Agentic | LiveBench | 28.3%† | 42.8%† |
| LiveBench Browser | LiveBench | 56.0%† | 16.0%† |
| LiveBench Agentic Coding | LiveBench | 58.3% | 62.2%† |
| Terminal-Bench 2.1 (Vals) | Vals AI | 77.5% | 80.9%† |
| MCP Atlas | Scale AI SEAL | – | 82.3% |
| HiL-Bench | Scale AI SEAL | – | 42.7%† |
| τ³-bench Banking | τ²-bench | – | 37.1% |
| τ³-Banking (AA) | Artificial Analysis | 32.8% | 46.0% |
| ITBench SRE (AA) | Artificial Analysis | – | 47.7% |
| Analyst Agent (AA) | Artificial Analysis | 60.0% | 38.8% |
| APEX-Agents (AA) | Artificial Analysis | – | 41.3% |
| CyberBench | Vals AI | 43.8% | 75.2% |
| Finance Agent v2 | Vals AI | 59.0% | 53.1% |
| Finance Agent v2.1 | Vals AI | – | 60.0% |
| Harvey's Legal Agent Benchmark v1.1 | Vals AI | – | 12.9% |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 8.8% | 12.9% |
| Legal Research Bench | Vals AI | 34.6% | 46.1% |
| SkillsBench | Vals AI | 65.9% | – |
| SREBench | Vals AI | 4.6% | – |
| Tax Agent Bench | Vals AI | 22.8% | 33.4% |
| Terminal-Bench 4.0 (Vals) | Vals AI | 12.1% | 17.2% |
| Terminal-Bench Science | Vals AI | 7.1% | 1.4% |
| Time Horizon Index: KSP | Vals AI | – | 10.5% |
| PostTrainBench | Epoch AI Benchmarking Hub | – | 32.0%† |
| GBAEval | Epoch AI Benchmarking Hub | – | 48.4%† |
| Vending-Bench 2 | Andon Labs | – | 5165.0† |
| AutomationBench | Artificial Analysis | 62.0% | 58.3% |
| GDP.pdf (AA) | Artificial Analysis | 23.6% | 22.0% |
| Harvey LAB-AA v1.1 | Harvey | – | 5.3% |
| EnterpriseOps-Gym | Artificial Analysis | 50.4%† | 45.3% |
| AA-Briefcase v1.1 | Artificial Analysis | – | 1499 |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | – | 8.1% |
| CWE-Bench (AA) | Artificial Analysis | – | 53.3% |
| DeepSec-Bench (AA) | Artificial Analysis | – | 12.4% |
| CyberGym end-to-end (AA) | Artificial Analysis | – | 58.0% |
| FACTS Search | 80.9%† | – | |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 71.6% | 72.2% |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 36.6% | 39.0% |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 97.2% | 97.2% |
| ProofBench | Vals AI | 58.0%† | 87.0%† |
| LiveBench Mathematics | LiveBench | 93.5% | 84.4%† |
| AIME 2026 | MathArena | – | 96.7%† |
| HMMT February 2026 | MathArena | – | 97.0%† |
| MathArena Apex | MathArena | – | 65.6%† |
| Arena Maths | Arena (LMArena) | 1505 | 1499 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 69.2% | 50.6% |
| LiveBench 2.0 Data Analysis | LiveBench | 68.0%† | 78.7%† |
| LiveBench Research | LiveBench | 76.5%† | 79.0%† |
| LiveBench 1.0 Data Analysis | LiveBench | 68.0% | 78.7%† |
| AA-Omniscience | Artificial Analysis | 26.5 | 19.7 |
| MMLU-Pro | Vals AI | 90.1% | 88.0% |
| LegalBench | Vals AI | 87.3% | 86.2% |
| CorpFin | Vals AI | – | 71.6%† |
| TaxEval | Vals AI | 74.7% | 75.7%† |
| BioMysteryBench | Vals AI | – | 72.6% |
| Excel Modeling Benchmark | Vals AI | 71.3% | 66.7% |
| Excel Modeling Benchmark v1.1 | Vals AI | – | 71.3% |
| MedCode | Vals AI | 53.4% | 49.4% |
| MedScribe | Vals AI | 83.9% | 88.0% |
| Public Benefits Bench v1.1 | Vals AI | – | 68.2% |
| GDP.pdf | Epoch AI Benchmarking Hub | 23.8% | 19.0% |
| AA-Omniscience: accuracy | Artificial Analysis | 55.3% | 47.6% |
| AA-Omniscience: non-hallucination | Artificial Analysis | 35.5% | 46.8% |
| Instruction following | |||
| LiveBench Language | LiveBench | 85.5% | 85.5%† |
| LiveBench Instruction Following | LiveBench | 79.9% | 71.4%† |
| Arena Instruction Following | Arena (LMArena) | 1487 | 1489 |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1490 | 1487 |
| EQ-Bench 4 | EQ-Bench | – | 1318† |
| Arena Creative Writing | Arena (LMArena) | 1492 | 1462 |
| Arena Multi-turn | Arena (LMArena) | 1489 | 1502 |
| Arena Longer Queries | Arena (LMArena) | 1501 | 1503 |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1314 | – |
| LiveBench Design | LiveBench | 73.2%† | 77.2%† |
| MMMU-Pro | Artificial Analysis | 85.5% | 80.5% |
| MMMU-Pro (Vals) | Vals AI | 89.0% | 88.2%† |
| MortgageTax | Vals AI | 66.7% | 66.3%† |
| SAGE | Vals AI | 49.2% | 52.8% |
| Blueprint-Bench 2 | Epoch AI Benchmarking Hub | – | 29.5%† |
| AA-3D | Artificial Analysis | – | 982 |
| Long context | |||
| AA-LCR | Artificial Analysis | 81.7% | 88.7% |
| MLCR | Artificial Analysis | 12.2% | 21.7% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 158.0† | 157.9† |
| LiveBench 2.0 | LiveBench | 67.2%† | 63.4%† |
| LiveBench 1.0 | LiveBench | 78.8% | 79.2%† |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 39.1 | 43.6 |
| Vals Index | Vals AI | – | 53.0 |
| Vals Finance Index | Vals AI | – | 58.0% |
| Vals RSI Index | Vals AI | 18.3% | 20.9% |
| Vals Multimodal Index | Vals AI | – | 73.4%† |
| Vals Router Index | Vals AI | – | 53.0%† |
| AA Cyber Index | Artificial Analysis | – | 41.3 |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | – | 26.6% |
| Metadata | |||
| AA Openness Index | Artificial Analysis | – | 38.9 |
Data as of 2026-10-25. Best configuration of each model; every score links to its source on the model pages.
What changed
- Gemini 3.7 Flash (high): 2 new results (Harvey's Legal Agent Benchmark v1, Vals RSI Index).
- Kimi K3 (max): 7 new results (Code Migration v1.1, Excel Modeling Benchmark v1.1, Finance Agent v2.1…).
- Gemini 3.7 Flash (high): input price moved from $0.38/M to $0.75/M (97% up); output price moved from $1.88/M to $3.75/M (99% up).
- Gemini 3.7 Flash (high): 1 score revision; input price moved from $0.75/M to $0.38/M (49% down); output price moved from $3.75/M to $1.88/M (50% down).
- Kimi K3 (max): 1 new result (Terminal-Bench-Science 0.1 (AA)); 1 score revision.
- Gemini 3.7 Flash (high): 1 new result (ForecastBench).
- Kimi K3 (max): 3 score revisions.
- Gemini 3.7 Flash (high): BenchLeader Index 62.9 → 63.5.
- Kimi K3 (max): 1 score revision.
- Gemini 3.7 Flash (high): BenchLeader Index 60.8 → 62.9.
Related comparisons
- Gemini 3.7 Flash vs Claude Opus 5.5
- Gemini 3.7 Flash vs Gemini 4 Argon
- Gemini 3.7 Flash vs Muse Wonder Full
- Kimi K3 vs Claude Opus 5.5
- Kimi K3 vs Gemini 4 Argon
- Kimi K3 vs Muse Wonder Full
- Kimi K3 vs Claude Fable 5.1
- Gemini 3.6 Flash vs Kimi K3
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
Gemini 3.7 Flash vs Kimi K3: questions
- Is Gemini 3.7 Flash better than Kimi K3?
- Gemini 3.7 Flash (high) and Kimi K3 (max) are neck and neck on quality (63.5 vs 64.8, within the ±4.8 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 2026-10-25, so check the category scores for your use.
- Which is cheaper, Gemini 3.7 Flash or Kimi K3, and will that change?
- Gemini 3.7 Flash is cheaper today: $1.50 against $6.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is Gemini 3.7 Flash better than Kimi K3 for coding?
- Kimi K3 scores higher in coding (65 vs 63 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1590 for Gemini 3.7 Flash against 1655 for Kimi K3.
- Is Gemini 3.7 Flash better than Kimi K3 for agents and tool use?
- Kimi K3 scores higher in agentic tasks (59 vs 48 on the category index, where 50 is average). On τ³-Banking (AA), the top shared agents & tools benchmark, it is 32.8% for Gemini 3.7 Flash against 46.0% for Kimi K3.
- Which is better for long documents?
- Kimi K3 scores higher on long-context benchmarks (70 vs 66). Both accept 1,048,576 tokens of context.
- Which has lower latency?
- Gemini 3.7 Flash (high) answers first: 2.5 s against 52.6 s to the first answer token, as Artificial Analysis measures it. Reasoning effort drives most of this: a higher setting thinks longer before it answers.
- Which is faster, Gemini 3.7 Flash or Kimi K3?
- Gemini 3.7 Flash streams faster: 88 against 41 output tokens per second.
- Do independent benchmarks agree on Gemini 3.7 Flash vs Kimi K3?
- Across the 59 benchmarks both have, Gemini 3.7 Flash leads 30, Kimi K3 leads 28 and 1 is a tie. Independent sources split: Gemini 3.7 Flash leads on Arena Text; Kimi K3 leads on AA Intelligence Index v4.3.2.
- When was Gemini 3.7 Flash released?
- Gemini 3.7 Flash was released on 13 Aug 2026.