GPT-6 Sol vs Kimi K3
Full profiles: GPT-6 Sol (OpenAI) and Kimi K3 (Moonshot AI): every benchmark, price and rank.
Updated · Data from 14 sources
As of 9 Nov 2026, GPT-6 Sol and Kimi K3 are neck and neck on quality (62.9 vs 62.0 on the BenchLeader Index, inside the ±4.5 margin), and GPT-6 Sol is 1.5× cheaper at $4.00 vs $6.00 per 1M tokens blended. Choose GPT-6 Sol for reasoning (66 vs 60), maths (69 vs 63), and coding (67 vs 65), where it also streams ~2× faster; choose Kimi K3 for human preference (68 vs 64), long context (69 vs 67), and multimodal (63 vs 61).
Key facts (as of 9 Nov 2026)
- GPT-6 Sol scores 62.9 on the BenchLeader Index and Kimi K3 scores 62.0 (BenchLeader, 9 Nov 2026).
- GPT-6 Sol costs $4.00 per 1M tokens blended (list price, 9 Nov 2026).
- Kimi K3 costs $6.00 per 1M tokens blended (list price, 9 Nov 2026).
- GPT-6 Sol streams ~2× faster than Kimi K3 (Artificial Analysis, 7-day median, 9 Nov 2026).
- Of the 60 benchmarks both models have, GPT-6 Sol leads 40 (BenchLeader, 9 Nov 2026).
- On Arena WebDev, GPT-6 Sol is rated 1674 and Kimi K3 1657 (LMArena, as published by 9 Nov 2026).
GPT-6 Sol benchmarks 
Kimi K3 benchmarks Which 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 | GPT-6 Sol | 67 vs 65 category score; Arena WebDev 1674 vs 1657 | Medium |
| Agents & tool use | Either | 60 vs 59 category score; Arena Agent 7 vs 0.6 | Either |
| Maths | GPT-6 Sol | 69 vs 63 category score; FrontierMath Tiers 1–3 89.8% vs 72.2% | Medium |
| Reasoning | GPT-6 Sol | 66 vs 60 category score; GPQA Diamond 94.3% vs 93.1% | Medium |
| Knowledge | GPT-6 Sol | 62 vs 60 category score; SimpleQA Verified 60.7% vs 50.6% | Medium |
| Long documents | Kimi K3 | 69 vs 67 category score; AA-LCR v1.1 88.7% vs 83.7% | Medium |
| Images and other media | Kimi K3 | 63 vs 61 category score; MMMU-Pro 80.5% vs 83.0% | Medium |
| Chat quality (human preference) | Kimi K3 | 68 vs 64 on Arena votes | Medium |
| Lowest cost | GPT-6 Sol | $4.00 vs $6.00 per 1M blended; agentic coding session $0.160 vs $0.240 | – |
| Fastest first answer | Either | 72.1 s (GPT-6 Sol (max)) vs 65.8 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
- GPT-6 Sol (max) and Kimi K3 (max) are neck and neck on quality (62.9 vs 62.0, within the ±4.5 margin).
- GPT-6 Sol (max) is stronger in agents & tools, coding, knowledge, maths, reasoning.
- Kimi K3 (max) is stronger in human preference, long context, multimodal.
- GPT-6 Sol (max) is 1.5× cheaper ($4.00 vs $6.00 per 1M blended).
- GPT-6 Sol (max) streams 2.9× faster (121 vs 42 tokens per second).
At a glance
| Spec | GPT-6 Sol | Kimi K3 |
|---|---|---|
| Developer | OpenAI | Moonshot AI |
| Released | 22 Sep 2026 | 16 Jul 2026 |
| Context window (tokens) | 1,050,000 | 1,048,576 |
| Compared at | GPT-6 Sol (max) | Kimi K3 (max) |
| Open weights | No | Yes |
| Price in / out / cached input per 1M | $2.00 / $10.00 / $0.200 | $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 | GPT-6 Sol (max) | Kimi K3 (max) | Leader |
|---|---|---|---|
| Reasoning | 66.2 | 59.5 | GPT-6 Sol |
| Coding | 66.5 | 64.5 | GPT-6 Sol |
| Agents & tools | 59.6 | 58.7 | level |
| Maths | 68.5 | 63.2 | GPT-6 Sol |
| Knowledge | 61.5 | 60.0 | GPT-6 Sol |
| Human preference | 64.2 | 68.0 | Kimi K3 |
| Multimodal | 60.7 | 63.0 | Kimi K3 |
| Long context | 66.6 | 69.2 | Kimi K3 |
On one chart
Each domain on the index's own scale, where 50 is an average model. Click a domain to see the benchmarks behind it. Pick up to 5 models from the list.
Not drawn, measured for fewer than two of these models: Instruction following.
ModelIndex
Where sources disagree
Across the 60 benchmarks both have, GPT-6 Sol leads 40, Kimi K3 leads 20. Independent sources split: GPT-6 Sol leads on AA Intelligence Index v5.0; Kimi K3 leads on Arena Text.
| Overall index | GPT-6 Sol (max) | Kimi K3 (max) | Leader |
|---|---|---|---|
| BenchLeader Index | 62.9 | 62.0 | neck and neck |
| AA Intelligence Index v5.0 | 44.2 | 17.4† | GPT-6 Sol |
| Arena Text | 1457 | 1487 | Kimi K3 |
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. Cost, tokens and time per benchmark task are measured: Artificial Analysis’ averages over its whole Intelligence Index run.
| Measure | GPT-6 Sol (max) | Kimi K3 (max) |
|---|---|---|
| Blended price per 1M tokens | $4.00 | $6.00 |
| Input / output per 1M | $2.00 / $10.00 | $3.00 / $15.00 |
| Cached input per 1M | $0.200 | $0.300 |
| Price per task: chat reply | $0.0038 | $0.0057 |
| Price per task: summarize a 30-page report | $0.030 | $0.045 |
| Price per task: code edit | $0.027 | $0.041 |
| Price per task: agentic coding session | $0.160 | $0.240 |
| Price per task: structured extraction | $0.0060 | $0.0090 |
| Agentic coding session, input mostly cached | $0.079 | $0.118 |
| Cost per benchmark task (AA Index) | $2.47 | – |
| Output tokens per task (AA Index) | 54k | – |
| Of which reasoning | 35k | – |
| Time per task (AA Index) | 7 min | – |
| Output speed | 121 tok/s | 42 tok/s |
| Time to first answer | 72.1 s | 65.8 s |
| Time to a complete answer | 130.2 s | 140.2 s |
Each model is compared at its best-ranked reasoning-effort setting. GPT-6 Sol (max) scores 62.9 and takes 72 s to its first answer token; GPT-6 Sol (xhigh) scores 58.6 in 24 s, GPT-6 Sol (high) scores 58.2 in 9 s. Kimi K3 (max) scores 62.0 and takes 66 s to its first answer token; Kimi K3 scores 59.5 in 1 s, Kimi K3 (low) scores 53.7 in 12 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
153 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 | GPT-6 Sol (max) | Kimi K3 (max) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 62.9 | 62.0 |
| Agents & tools score | BenchLeader | 59.6 | 58.7 |
| Coding score | BenchLeader | 66.5 | 64.5 |
| Human preference score | BenchLeader | 64.2 | 68.0 |
| Knowledge score | BenchLeader | 61.5 | 60.0 |
| Long context score | BenchLeader | 66.6 | 69.2 |
| Maths score | BenchLeader | 68.5 | 63.2 |
| Multimodal score | BenchLeader | 60.7 | 63.0 |
| Reasoning score | BenchLeader | 66.2 | 59.5 |
| Blended price $/M | BenchLeader | $4.00 | $6.00 |
| Output speed | BenchLeader | 121 tok/s | 42 tok/s |
| Time to first answer | BenchLeader | 72.1 s | 65.8 s |
| Context window | BenchLeader | 1,050,000 tokens | 1,048,576 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 94.3% | 93.1% |
| SimpleBench | SimpleBench | – | 60.7% |
| Arena Hard Prompts | Arena (LMArena) | 1483 | 1515 |
| LiveBench 2.0 Reasoning | LiveBench | – | 90.7%† |
| LiveBench 1.0 Reasoning | LiveBench | 88.7% | 90.7%† |
| GPQA Diamond (AA) | Artificial Analysis | – | 93.5% |
| Humanity's Last Exam (AA) | Artificial Analysis | 47.9% | 46.9% |
| HLE-Diamond (AA) | Artificial Analysis | 45.2% | 26.9% |
| GPQA Diamond (Vals) | Vals AI | – | 92.9%† |
| ARC-AGI-1 | ARC Prize | 95.5% | 94.5% |
| ARC-AGI-2 | ARC Prize | 89.6% | 60.4% |
| ARC-AGI-3 | ARC Prize | 23.0% | – |
| CritPt | Artificial Analysis | 30.9% | 23.4% |
| CritPt v2 hard subset (AA) | Artificial Analysis | 66.5% | 40.5% |
| MysteryMechanism | Vals AI | 30.2% | – |
| Chess Puzzles | Epoch AI Benchmarking Hub | 56.0% | 36.0% |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 56.0% | 26.0% |
| LMCA | Epoch AI Benchmarking Hub | 59.1% | 52.7% |
| DTBench | Epoch AI Benchmarking Hub | 97.3% | 91.2% |
| ForecastBench | ForecastBench | – | 64.7% |
| Coding | |||
| SciCode | SciCode | 57.6% | 59.5% |
| WeirdML | WeirdML | – | 82.6% |
| FrontierCode | Cognition | 49.3% | 44.2%† |
| Arena Coding | Arena (LMArena) | 1520 | 1537 |
| Arena WebDev | Arena (LMArena) | 1674 | 1657 |
Show all 153 benchmarks
| Metric | Publisher | GPT-6 Sol (max) | Kimi K3 (max) |
|---|---|---|---|
| Coding | |||
| LiveBench 2.0 Coding | LiveBench | – | 53.1%† |
| LiveBench 1.0 Coding | LiveBench | 81.8% | 81.5%† |
| LiveCodeBench | Vals AI | – | 87.2% |
| IOI | Vals AI | 82.6% | 48.9% |
| SWE-bench (Vals) | Vals AI | – | 93.4%† |
| Code Migration | Vals AI | 57.2% | 16.1% |
| Code Migration v1.1 | Vals AI | – | 28.6% |
| ProgramBench | Vals AI | 2.0% | 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 | 87.8% | 85.0% |
| Surface Evolver Bench | Epoch AI Benchmarking Hub | – | 93.0% |
| DeepSWE v1.1 | Epoch AI Benchmarking Hub | – | 68.5% |
| ALE-Bench | Epoch AI Benchmarking Hub | 2462 | 1524.5 |
| FrontierSWE | Epoch AI Benchmarking Hub | – | 25.9% |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 43.9% | 12.6% |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | – | 85.0% |
| Startrise Frontend | Startrise | – | 79.8%† |
| GameBench 2 Overall | DuelLab GameBench 2 | 72.9† | 47.2† |
| GameBench 2 by reasoning setting | DuelLab GameBench 2 | 79.3† | 43.8 |
| Design Arena Website | Design Arena | 1285† | 1332† |
| FrontierCode 1.1 (AA) | Artificial Analysis | – | 43.6% |
| FrontierCode 2.0 (AA) | Artificial Analysis | 35.6% | 10.9%† |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 50.3% | – |
| APEX-Agents | Mercor | 54.3% | 47.8%† |
| Arena Agent | Arena (LMArena) | 7 | 0.6 |
| LiveBench Long Agentic | LiveBench | – | 36.0%† |
| LiveBench Browser | LiveBench | – | 75.2%† |
| LiveBench Agentic Coding | LiveBench | 52.9% | 62.2%† |
| Terminal-Bench 2.1 (Vals) | Vals AI | 83.2% | 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 | – | 46.0% |
| ITBench SRE (AA) | Artificial Analysis | 49.4% | 47.7% |
| Analyst Agent (AA) | Artificial Analysis | – | 38.8% |
| APEX-Agents (AA) | Artificial Analysis | – | 41.3% |
| CyberBench | Vals AI | 78.0% | 75.2% |
| Finance Agent v2 | Vals AI | 49.0% | 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 | 1.7% | 12.9% |
| Legal Research Bench | Vals AI | 28.9% | 46.1% |
| SREBench | Vals AI | 30.5% | – |
| Tax Agent Bench | Vals AI | 15.0% | 33.4% |
| Terminal-Bench 4.0 (Vals) | Vals AI | 44.4% | 17.2% |
| Terminal-Bench Science | Vals AI | 30.0% | 1.4% |
| Time Horizon Index: KSP | Vals AI | – | 10.5% |
| Arena Search | Arena (LMArena) | 1229 | – |
| PostTrainBench | Epoch AI Benchmarking Hub | – | 32.0%† |
| GBAEval | Epoch AI Benchmarking Hub | – | 48.4%† |
| Vending-Bench 2 | Andon Labs | 14427.9† | 5165.0† |
| AutomationBench | Artificial Analysis | 61.6% | 58.3% |
| GDP.pdf (AA) | Artificial Analysis | 25.2% | 22.0% |
| Harvey LAB-AA v1.1 | Harvey | 3.6% | 5.3% |
| EnterpriseOps-Gym | Artificial Analysis | – | 45.3% |
| AA-Briefcase v1.1 | Artificial Analysis | 1476 | 1497 |
| GDPx | GDPx by SemiAnalysis | – | 3.7% |
| Design Arena Agentic | Design Arena | 1212† | 1257† |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 30.0% | 8.1% |
| CWE-Bench (AA) | Artificial Analysis | 64.2% | 53.3% |
| DeepSec-Bench (AA) | Artificial Analysis | 46.3% | 12.4% |
| CyberGym end-to-end (AA) | Artificial Analysis | 0.0% | 58.0% |
| FACTS Search | – | 90.3%† | |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 89.8% | 72.2% |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 90.0% | 39.0% |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 100.0% | 97.2% |
| ProofBench | Vals AI | 83.0%† | 87.0%† |
| LiveBench Mathematics | LiveBench | 96.4% | 84.4%† |
| AIME 2026 | MathArena | – | 96.7%† |
| HMMT February 2026 | MathArena | – | 97.0%† |
| MathArena Apex | MathArena | – | 65.6%† |
| Arena Maths | Arena (LMArena) | 1440 | 1502 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 60.7% | 50.6% |
| LiveBench 2.0 Data Analysis | LiveBench | – | 78.7%† |
| LiveBench Research | LiveBench | – | 79.8%† |
| LiveBench 1.0 Data Analysis | LiveBench | 81.2% | 78.7%† |
| AA-Omniscience v1.1 | Artificial Analysis | 21.4 | -1.7† |
| MMLU-Pro | Vals AI | – | 88.0% |
| LegalBench | Vals AI | – | 86.2% |
| CorpFin | Vals AI | – | 71.6%† |
| TaxEval | Vals AI | – | 75.7%† |
| PRBench Legal | Scale AI SEAL | 38.6%† | – |
| BioMysteryBench | Vals AI | 74.8% | 72.6% |
| Excel Modeling Benchmark | Vals AI | 71.5% | 66.7% |
| Excel Modeling Benchmark v1.1 | Vals AI | – | 71.3% |
| MedCode | Vals AI | 47.1% | 49.4% |
| MedScribe | Vals AI | 82.0% | 88.0% |
| Public Benefits Bench v1.1 | Vals AI | 56.6% | 68.2% |
| EBR-bench | Epoch AI Benchmarking Hub | 53.3% | – |
| GDP.pdf | Epoch AI Benchmarking Hub | 26.4% | 19.0% |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 51.1% | 42.4%† |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 39.1% | 23.4%† |
| HealthBench Professional (AA) | OpenAI | 62.3% | 53.2% |
| Instruction following | |||
| LiveBench Language | LiveBench | 85.3% | 85.5%† |
| LiveBench Instruction Following | LiveBench | 68.6% | 71.4%† |
| Arena Instruction Following | Arena (LMArena) | 1448 | 1489 |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1457 | 1487 |
| EQ-Bench 4 | EQ-Bench | – | 1319† |
| Arena Creative Writing | Arena (LMArena) | 1430 | 1459 |
| Arena Multi-turn | Arena (LMArena) | 1466 | 1501 |
| Arena Longer Queries | Arena (LMArena) | 1466 | 1500 |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1232 | – |
| LiveBench Design | LiveBench | – | 72.5%† |
| MMMU-Pro | Artificial Analysis | 83.0% | 80.5% |
| MMMU-Pro (Vals) | Vals AI | – | 88.2%† |
| MortgageTax | Vals AI | – | 66.3%† |
| SAGE | Vals AI | – | 86.8% |
| Blueprint-Bench 2 | Epoch AI Benchmarking Hub | 36.9%† | 29.5%† |
| Code2Video-AA | Artificial Analysis | – | 1041 |
| AA-3D | Artificial Analysis | 1131 | 984 |
| Long context | |||
| AA-LCR v1.1 | Artificial Analysis | 83.7% | 88.7% |
| MLCR | Artificial Analysis | 33.9% | 21.7% |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 164.1† | 158.5† |
| LiveBench 2.0 | LiveBench | – | 65.0%† |
| LiveBench 1.0 | LiveBench | 79.3% | 79.2%† |
| AA Intelligence Index v5.0 | Artificial Analysis | 44.2 | 17.4† |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 47.6 | 43.6 |
| AA Progress Index v1 | Artificial Analysis | 334 | 319.9 |
| Vals Index | Vals AI | – | 53.0 |
| Vals Finance Index | Vals AI | – | 58.0% |
| Vals RSI Index | Vals AI | 28.1% | 20.9% |
| Vals Science Index | Vals AI | 45.0% | – |
| Vals Multimodal Index | Vals AI | – | 73.4%† |
| Vals Router Index | Vals AI | – | 53.0%† |
| AA Cyber Index | Artificial Analysis | 36.8 | 41.3 |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | 13.5%† | 26.6% |
| Metadata | |||
| AA Openness Index | Artificial Analysis | – | 38.9 |
Data as of 2026-11-09. Best configuration of each model; every score links to its source on the model pages.
DuelLab GameBench 2, release gb2-20261103-eaea31df94e6 (https://benchmarks.duellab.org/releases/gb2-20261103-eaea31df94e6/), CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Scores mapped to BenchLeader model names; not endorsed by DuelLab.
What changed
- GPT-6 Sol (max): BenchLeader Index 63.5 → 62.9.
- GPT-6 Sol (max): 2 score revisions.
- GPT-6 Sol (max): 1 score revision.
- GPT-6 Sol (max): 1 score revision.
- GPT-6 Sol (max): 6 new results (AA Intelligence Index v5.0, AA-Omniscience v1.1, CritPt v2 hard subset (AA)…).
- Kimi K3 (max): 2 score revisions.
- GPT-6 Sol (max): 1 score revision.
- Kimi K3 (max): 1 new result (MirrorCode).
- GPT-6 Sol (max): 1 score revision.
- Kimi K3 (max): 1 score revision.
Related comparisons
- GPT-6 Sol vs Claude Fable 5.2
- GPT-6 Sol vs Claude Opus 5.5
- GPT-6 Sol vs Gemini 4 Carbon
- Kimi K3 vs Claude Fable 5.2
- Kimi K3 vs Claude Opus 5.5
- Kimi K3 vs Gemini 4 Carbon
- Kimi K3 vs GPT-6 Astra
- GPT-6 Sol vs Kimi K2.6
Rankings: Best LLM overall · Best LLM for agents · Best cheap LLMs
GPT-6 Sol vs Kimi K3: questions
- Is GPT-6 Sol better than Kimi K3?
- GPT-6 Sol (max) and Kimi K3 (max) are neck and neck on quality (62.9 vs 62.0, within the ±4.5 margin). The BenchLeader Index combines every independent quality benchmark; the two are within the margin of error as of 9 Nov 2026, so check the category scores for your use.
- GPT-6 Sol vs Kimi K3: which is cheaper, and will that change?
- GPT-6 Sol is cheaper today: $4.00 against $6.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is GPT-6 Sol or Kimi K3 better for coding?
- GPT-6 Sol scores higher in coding (67 vs 65 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1674 for GPT-6 Sol against 1657 for Kimi K3.
- Which is faster, GPT-6 Sol or Kimi K3?
- GPT-6 Sol streams ~2× faster (121 vs 42 output tokens per second, Artificial Analysis 7-day median).
- Is GPT-6 Sol or Kimi K3 better for agents and tool use?
- They are level in agentic tasks (60 vs 59 on the category index). On Arena Agent, the top shared agents & tools benchmark, it is 7 for GPT-6 Sol against 0.6 for Kimi K3.
More questions (4)
- Which is better for long documents?
- Kimi K3 scores higher on long-context benchmarks (69 vs 67). Both take about 1.1M tokens: GPT-6 Sol 1,050,000, Kimi K3 1,048,576.
- Which has lower latency?
- Kimi K3 answers first: 65.8 s against 72.1 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.
- Do independent benchmarks agree on GPT-6 Sol vs Kimi K3?
- Across the 60 benchmarks both have, GPT-6 Sol leads 40, Kimi K3 leads 20. Independent sources split: GPT-6 Sol leads on AA Intelligence Index v5.0; Kimi K3 leads on Arena Text.
- When was GPT-6 Sol released?
- GPT-6 Sol was released on 22 Sep 2026.