GPT-6.1 Sol vs Kimi K2.5
Full profiles: GPT-6.1 Sol (OpenAI) and Kimi K2.5 (Moonshot AI): every benchmark, price and rank.
Updated · Data from 12 sources
As of 29 Oct 2026, GPT-6.1 Sol is ahead of Kimi K2.5 on quality (67.2 vs 58.8 on the BenchLeader Index), but Kimi K2.5 is 3.3× cheaper at $1.20 vs $4.00 per 1M tokens blended. GPT-6.1 Sol leads in every category we measure both in.
Key facts (as of 29 Oct 2026)
- GPT-6.1 Sol scores 67.2 on the BenchLeader Index and Kimi K2.5 scores 58.8 (BenchLeader, 29 Oct 2026).
- GPT-6.1 Sol costs $4.00 per 1M tokens blended (list price, 29 Oct 2026).
- Kimi K2.5 costs $1.20 per 1M tokens blended (list price, 29 Oct 2026).
- Of the 22 benchmarks both models have, GPT-6.1 Sol leads 22 (BenchLeader, 29 Oct 2026).
- On Arena WebDev, GPT-6.1 Sol is rated 1752 and Kimi K2.5 1437 (LMArena, as published by 29 Oct 2026).
GPT-6.1 Sol benchmarks max OpenAI
Kimi K2.5 benchmarks thinking 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 | GPT-6.1 Sol | 72 vs 56 category score; Arena WebDev 1752 vs 1437 | High |
| Agents & tool use | GPT-6.1 Sol | 64 vs 45 category score; Finance Agent v2 52.0% vs 35.8% | High |
| Maths | GPT-6.1 Sol | 73 vs 56 category score; Arena Maths 1513 vs 1471 | High |
| Reasoning | GPT-6.1 Sol | 73 vs 67 category score; Arena Hard Prompts 1504 vs 1471 | Medium |
| Knowledge | GPT-6.1 Sol | 73 vs 60 category score; Excel Modeling Benchmark 70.8% vs 28.5% | High |
| Long documents | GPT-6.1 Sol | 66 vs 64 category score; AA-LCR v1.1 83.0% vs 78.0% | Medium |
| Images and other media | GPT-6.1 Sol | 66 vs 59 category score; Arena Vision 1288 vs 1269 | Medium |
| Chat quality (human preference) | GPT-6.1 Sol | 67 vs 63 on Arena votes | Medium |
| Lowest cost | Kimi K2.5 | $1.20 vs $4.00 per 1M blended; agentic coding session $0.048 vs $0.160 | – |
| Fastest first answer | Kimi K2.5 | 36.0 s (Kimi K2.5 (thinking)) vs 249.5 s (GPT-6.1 Sol (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-6.1 Sol (max) leads on quality: 67.2 vs 58.8.
- GPT-6.1 Sol (max) is stronger in agents & tools, coding, human preference, knowledge, long context, maths, multimodal, reasoning.
- Kimi K2.5 (thinking) is stronger in instruction following.
- Kimi K2.5 (thinking) is 3.3× cheaper ($1.20 vs $4.00 per 1M blended).
At a glance
| Spec | GPT-6.1 Sol | Kimi K2.5 |
|---|---|---|
| Developer | OpenAI | Moonshot AI |
| Released | 29 Sept 2026 | 27 Jan 2026 |
| Context window (tokens) | 1,050,000 | 262,144 |
| Compared at | GPT-6.1 Sol (max) | Kimi K2.5 (thinking) |
| Open weights | No | Yes |
| Price in / out / cached input per 1M | $2.00 / $10.00 / $0.100 | $0.600 / $3.00 / $0.100 |
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-6.1 Sol (max) | Kimi K2.5 (thinking) | Leader |
|---|---|---|---|
| Reasoning | 73.2 | 67.3 | GPT-6.1 Sol |
| Coding | 71.6 | 55.7 | GPT-6.1 Sol |
| Agents & tools | 64.4 | 45.1 | GPT-6.1 Sol |
| Maths | 72.8 | 56.2 | GPT-6.1 Sol |
| Knowledge | 73.0 | 60.0 | GPT-6.1 Sol |
| Instruction following | – | 69.0 | not comparable |
| Human preference | 67.1 | 63.4 | GPT-6.1 Sol |
| Multimodal | 65.9 | 59.3 | GPT-6.1 Sol |
| Long context | 66.4 | 63.8 | GPT-6.1 Sol |
Where sources disagree
Across the 22 benchmarks both have, GPT-6.1 Sol leads 22, Kimi K2.5 leads 0. Every overall index that rates both puts GPT-6.1 Sol ahead.
| Overall index | GPT-6.1 Sol (max) | Kimi K2.5 (thinking) | Leader |
|---|---|---|---|
| BenchLeader Index | 67.2 | 58.8 | GPT-6.1 Sol |
| Arena Text | 1480 | 1450 | GPT-6.1 Sol |
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 | GPT-6.1 Sol (max) | Kimi K2.5 (thinking) |
|---|---|---|
| Blended price per 1M tokens | $4.00 | $1.20 |
| Input / output per 1M | $2.00 / $10.00 | $0.600 / $3.00 |
| Cached input per 1M | $0.100 | $0.100 |
| Price per task: chat reply | $0.0038 | $0.0011 |
| Price per task: summarize a 30-page report | $0.030 | $0.0090 |
| Price per task: code edit | $0.027 | $0.0081 |
| Price per task: agentic coding session | $0.160 | $0.048 |
| Price per task: structured extraction | $0.0060 | $0.0018 |
| Cost to run a full benchmark suite | $1.27 | – |
| Output speed | 79 tok/s | 87 tok/s |
| Time to first answer | 249.5 s | 36.0 s |
| Time to a complete answer | 255.8 s | 41.7 s |
Each model is compared at its best-ranked reasoning-effort setting. GPT-6.1 Sol (max) scores 67.2 and takes 249 s to its first answer token; GPT-6.1 Sol (xhigh) scores 64.3 in 97 s, GPT-6.1 Sol (high) scores 64.0 in 28 s. Kimi K2.5 (thinking) scores 58.8 and takes 36 s to its first answer token; Kimi K2.5 scores 54.3 in 1 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
159 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-6.1 Sol (max) | Kimi K2.5 (thinking) |
|---|---|---|---|
| BenchLeader Index | BenchLeader | 67.2 | 58.8 |
| Agents & tools score | BenchLeader | 64.4 | 45.1 |
| Coding score | BenchLeader | 71.6 | 55.7 |
| Human preference score | BenchLeader | 67.1 | 63.4 |
| Knowledge score | BenchLeader | 73.0 | 60.0 |
| Long context score | BenchLeader | 66.4 | 63.8 |
| Maths score | BenchLeader | 72.8 | 56.2 |
| Multimodal score | BenchLeader | 65.9 | 59.3 |
| Reasoning score | BenchLeader | 73.2 | 67.3 |
| Instruction following score | BenchLeader | – | 69.0 |
| Blended price $/M | BenchLeader | $4.00 | $1.20 |
| Output speed | BenchLeader | 79 tok/s | 87 tok/s |
| Time to first answer | BenchLeader | 249.5 s | 36.0 s |
| Context window | BenchLeader | 1,050,000 tokens | 262,144 tokens |
| Reasoning | |||
| GPQA Diamond | Epoch AI Benchmarking Hub | 95.4% | 87.6%† |
| Humanity's Last Exam | Scale AI / CAIS | – | 24.4%† |
| SimpleBench | SimpleBench | – | 46.8%† |
| Arena Hard Prompts | Arena (LMArena) | 1504 | 1471 |
| LiveBench 2.0 Reasoning | LiveBench | 92.6%† | – |
| LiveBench 1.0 Reasoning | LiveBench | 92.6% | – |
| GPQA Diamond (AA) | Artificial Analysis | – | 87.9% |
| Humanity's Last Exam (AA) | Artificial Analysis | 52.9% | 30.7% |
| HLE-Diamond (AA) | Artificial Analysis | 58.8% | – |
| GPQA Diamond (Vals) | Vals AI | – | 84.1% |
| Kagi LLM Benchmark | Kagi LLM Benchmark | – | 78.5% |
| ARC-AGI-1 | ARC Prize | 96.5% | 65.3%† |
| ARC-AGI-2 | ARC Prize | 94.2% | 11.8%† |
| ARC-AGI-3 | ARC Prize | 96.2% | – |
| CritPt | Artificial Analysis | 31.7% | 3.1% |
| CritPt v2 hard subset (AA) | Artificial Analysis | 68.5% | – |
| MysteryMechanism | Vals AI | 46.4% | – |
| Chess Puzzles | Epoch AI Benchmarking Hub | 68.0% | – |
| Mystery Game Puzzles | Epoch AI Benchmarking Hub | 80.0% | – |
| CL-bench | Epoch AI Benchmarking Hub | – | 19.3%† |
| CL-bench Life | Epoch AI Benchmarking Hub | – | 13.2%† |
| LMCA | Epoch AI Benchmarking Hub | 62.8% | – |
| DTBench | Epoch AI Benchmarking Hub | 96.5% | – |
| Coding | |||
| SWE-bench Verified (Epoch) | Epoch AI Benchmarking Hub | – | 73.8%† |
| SciCode | SciCode | 54.2% | 49.0%† |
Show all 159 benchmarks
| Metric | Publisher | GPT-6.1 Sol (max) | Kimi K2.5 (thinking) |
|---|---|---|---|
| Coding | |||
| WeirdML | WeirdML | – | 45.6%† |
| FrontierCode | Cognition | 50.2%† | – |
| Arena Coding | Arena (LMArena) | 1537 | 1502 |
| Arena WebDev | Arena (LMArena) | 1752 | 1437 |
| LiveBench 2.0 Coding | LiveBench | 96.0%† | – |
| LiveBench 1.0 Coding | LiveBench | 80.4% | – |
| SciCode (AA) | Artificial Analysis | 54.2% | – |
| LiveCodeBench | Vals AI | – | 83.9% |
| IOI | Vals AI | 96.9% | – |
| SWE-bench (Vals) | Vals AI | – | 70.0% |
| SWE-bench Verified (bash only) | SWE-bench | – | 70.8%† |
| SWE-bench Verified (any scaffold) | SWE-bench | – | 70.8%† |
| Code Migration | Vals AI | 65.1% | 7.0% |
| Code Migration v1.1 | Vals AI | 77.3% | – |
| ProgramBench | Vals AI | 3.0% | – |
| SWE-bench Multimodal | Vals AI | 43.5% | – |
| Vibe Code Bench v1.1 | Vals AI | 88.9% | 17.5% |
| SWE Atlas: Codebase QnA | Scale AI SEAL | – | 13.1%† |
| SWE Atlas: Refactoring | Scale AI SEAL | – | 20.9%† |
| SWE Atlas: Test Writing | Scale AI SEAL | – | 25.8%† |
| ALE-Bench | Epoch AI Benchmarking Hub | 2743.8 | 821.6† |
| Terminal-Bench 4.0 (AA) | Artificial Analysis | 56.1% | – |
| Terminal-Bench 2.1 (AA) | Artificial Analysis | – | 45.7% |
| Startrise Frontend | Startrise | 78.6%† | – |
| FrontierCode 1.1 (AA) | Artificial Analysis | 49.1% | – |
| FrontierCode 2.0 (AA) | Artificial Analysis | 40.6% | – |
| Agents & tools | |||
| Terminal-Bench | Terminal-Bench | 58.2% | 43.2%† |
| APEX-Agents | Mercor | 60.0% | – |
| Arena Agent | Arena (LMArena) | 13.6 | – |
| LiveBench Long Agentic | LiveBench | 68.7%† | – |
| LiveBench Browser | LiveBench | 93.3%† | – |
| LiveBench Agentic Coding | LiveBench | 54.5% | – |
| Terminal-Bench Hard | Artificial Analysis | – | 34.9% |
| τ²-Bench Telecom (AA) | Artificial Analysis | – | 95.9% |
| Terminal-Bench 2.1 (Vals) | Vals AI | – | 42.0% |
| MCP Atlas | Scale AI SEAL | – | 64.4%† |
| HiL-Bench | Scale AI SEAL | 62.3%† | – |
| GDPval-AA v2.1, normalised score | Artificial Analysis | 54.8% | – |
| GDPval-AA v2.1 | Artificial Analysis | 1608 | – |
| τ³-Banking (AA) | Artificial Analysis | – | 14.2% |
| Analyst Agent (AA) | Artificial Analysis | 50.0% | – |
| APEX-Agents (AA) | Artificial Analysis | – | 11.5% |
| CyberBench | Vals AI | 39.3% | – |
| Finance Agent v2 | Vals AI | 52.0% | 35.8% |
| Finance Agent v2.1 | Vals AI | 61.8% | – |
| Harvey's Legal Agent Benchmark v1.1 | Vals AI | 6.3% | – |
| Harvey's Legal Agent Benchmark v1 | Vals AI | 5.4% | 0.0% |
| Legal Research Bench | Vals AI | 38.5% | 15.9% |
| Robo-bench | Vals AI | 28.3% | – |
| SREBench | Vals AI | 50.8% | – |
| Tax Agent Bench | Vals AI | 21.8% | – |
| Terminal-Bench 2.0 (Vals) | Vals AI | – | 40.5% |
| Terminal-Bench 4.0 (Vals) | Vals AI | 55.0% | – |
| Terminal-Bench Science | Vals AI | 52.9% | – |
| Arena Search | Arena (LMArena) | 1250 | – |
| Vending-Bench 2 | Andon Labs | – | 1198.5† |
| AutomationBench | Artificial Analysis | 64.9% | – |
| GDP.pdf (AA) | Artificial Analysis | 31.0% | – |
| Harvey LAB-AA v1.1 | Harvey | 6.9% | – |
| AA-Briefcase v1.1 | Artificial Analysis | 1555 | – |
| Terminal-Bench-Science 0.1 (AA) | Artificial Analysis | 58.1% | – |
| FACTS Search | 88.2%† | – | |
| Maths | |||
| FrontierMath Tiers 1–3 | Epoch AI Benchmarking Hub | 93.7% | – |
| FrontierMath Tier 4 | Epoch AI Benchmarking Hub | 100.0% | – |
| OTIS Mock AIME | Epoch AI Benchmarking Hub | 100.0% | 92.2%† |
| ProofBench | Vals AI | 99.0%† | – |
| LiveBench Mathematics | LiveBench | 96.8% | – |
| AIME (Vals) | Vals AI | – | 95.6% |
| AIME 2026 | MathArena | – | 95.8% |
| HMMT February 2026 | MathArena | – | 87.1% |
| MathArena Apex | MathArena | – | 8.8% |
| Arena Maths | Arena (LMArena) | 1513 | 1471 |
| Knowledge | |||
| SimpleQA Verified | Epoch AI Benchmarking Hub | 73.9% | 34.3%† |
| LiveBench 2.0 Data Analysis | LiveBench | 82.7%† | – |
| LiveBench Research | LiveBench | 83.5%† | – |
| LiveBench 1.0 Data Analysis | LiveBench | 82.7% | – |
| AA-Omniscience v1.1 | Artificial Analysis | 42.0 | – |
| MMLU-Pro | Vals AI | – | 85.9% |
| CorpFin | Vals AI | – | 68.3% |
| TaxEval | Vals AI | – | 74.2% |
| MedQA | Vals AI | – | 94.4% |
| PRBench Finance | Scale AI SEAL | 46.2%† | 46.5%† |
| PRBench Legal | Scale AI SEAL | 47.9%† | 43.8%† |
| MultiNRC | Scale AI SEAL | – | 35.2%† |
| BioMysteryBench | Vals AI | 79.6% | – |
| CaseLaw v2 | Vals AI | – | 58.7% |
| ClaimBench | Vals AI | 100.0% | – |
| Excel Modeling Benchmark | Vals AI | 70.8% | 28.5% |
| Excel Modeling Benchmark v1.1 | Vals AI | 71.1% | – |
| MedCode | Vals AI | 48.8% | 39.3% |
| MedScribe | Vals AI | 86.5% | 76.4% |
| Public Benefits Bench v1.1 | Vals AI | 59.3% | – |
| EBR-bench | Epoch AI Benchmarking Hub | 54.3% | – |
| GDP.pdf | Epoch AI Benchmarking Hub | 35.0% | – |
| AA-Omniscience v1.1: accuracy | Artificial Analysis | 61.8% | – |
| AA-Omniscience v1.1: non-hallucination | Artificial Analysis | 48.1% | – |
| HealthBench Professional (AA) | OpenAI | 65.1% | – |
| Instruction following | |||
| LiveBench Language | LiveBench | 90.1% | – |
| LiveBench Instruction Following | LiveBench | 74.2% | – |
| IFBench | Artificial Analysis | – | 70.2% |
| MultiChallenge | Scale AI SEAL | 82.0%† | 61.4%† |
| TutorBench | Scale AI SEAL | – | 54.6%† |
| Arena Instruction Following | Arena (LMArena) | 1485 | 1439 |
| Human preference | |||
| Arena Text | Arena (LMArena) | 1480 | 1450 |
| EQ-Bench 4 | EQ-Bench | 1272† | – |
| Arena Creative Writing | Arena (LMArena) | 1453 | 1423 |
| Arena Multi-turn | Arena (LMArena) | 1483 | 1452 |
| Arena Longer Queries | Arena (LMArena) | 1489 | 1458 |
| Multimodal | |||
| Arena Vision | Arena (LMArena) | 1288 | 1269 |
| LiveBench Design | LiveBench | 82.1%† | – |
| MMMU-Pro | Artificial Analysis | 86.0% | 75.4% |
| VISTA | Scale AI SEAL | – | 41.9%† |
| MMMU-Pro (Vals) | Vals AI | – | 84.3% |
| MortgageTax | Vals AI | – | 66.5% |
| SAGE | Vals AI | 90.5% | – |
| Code2Video-AA | Artificial Analysis | 1369 | – |
| AA-3D | Artificial Analysis | 1119 | – |
| Long context | |||
| Fiction.LiveBench 120k | Fiction.live | – | 78.1%† |
| AA-LCR v1.1 | Artificial Analysis | 83.0% | 78.0% |
| Arena Document | Arena (LMArena) | – | 1436 |
| MLCR | Artificial Analysis | 38.3% | – |
| Composite | |||
| Epoch Capabilities Index | Epoch AI Benchmarking Hub | 168.4† | 148.1† |
| LiveBench 2.0 | LiveBench | 84.1%† | – |
| LiveBench 1.0 | LiveBench | 81.6% | – |
| AA Intelligence Index v5.0 | Artificial Analysis | 54.5 | – |
| AA Intelligence Index v4.3.2 | Artificial Analysis | 51.8 | 23.5 |
| AA Progress Index v1 | Artificial Analysis | 348.5 | 270.5 |
| Vals Index | Vals AI | 63.9 | – |
| Vals Finance Index | Vals AI | 55.8% | – |
| Vals RSI Index | Vals AI | 26.6% | 6.8%† |
| Vals Science Index | Vals AI | 59.6% | – |
| Vals Router Index | Vals AI | 63.9%† | – |
| Safety & honesty | |||
| FORTRESS | Scale AI SEAL | – | 41.1%† |
| MASK | Scale AI SEAL | – | 70.5%† |
Data as of 2026-10-29. Best configuration of each model; every score links to its source on the model pages.
What changed
- GPT-6.1 Sol (max): 5 new results (AA Intelligence Index v5.0, AA Progress Index v1, HLE-Diamond (AA)…); 3 score revisions; cost per task moved from $0.72/M to $1.27/M (76% up).
- Kimi K2.5 (thinking): 1 new result (AA Progress Index v1).
- GPT-6.1 Sol (max): 10 new results (ClaimBench, Code Migration v1.1, Excel Modeling Benchmark v1.1…).
- Kimi K2.5 (thinking): 1 new result (Harvey's Legal Agent Benchmark v1).
- GPT-6.1 Sol (max): BenchLeader Index 67.9 → 67.4.
- GPT-6.1 Sol (max): 5 score revisions.
- Kimi K2.5 (thinking): output price moved from $2.75/M to $3.00/M (9% up).
- GPT-6.1 Sol (max): 2 score revisions.
- GPT-6.1 Sol (max): 2 new results (LMCA, DTBench).
- GPT-6.1 Sol (max): 2 score revisions.
Related comparisons
- GPT-6.1 Sol vs Claude Opus 5.5
- GPT-6.1 Sol vs GPT-6 Astra
- GPT-6.1 Sol vs Claude Fable 5.1
- Kimi K2.5 vs Claude Opus 5.5
- Kimi K2.5 vs GPT-6 Astra
- Kimi K2.5 vs Claude Fable 5.1
- GPT-6.1 Sol vs Gemini 4 Argon
- GPT-6.1 Sol vs Claude Opus 5
Rankings: Best LLM overall · Best LLM for coding · Best LLM for agents · Best cheap LLMs
GPT-6.1 Sol vs Kimi K2.5: questions
- Is GPT-6.1 Sol better than Kimi K2.5?
- GPT-6.1 Sol (max) leads on quality: 67.2 vs 58.8. The BenchLeader Index combines every independent quality benchmark; GPT-6.1 Sol (max) is ahead overall as of 29 Oct 2026, but check the category scores for your use.
- GPT-6.1 Sol vs Kimi K2.5: which is cheaper, and will that change?
- Kimi K2.5 is cheaper today: $1.20 against $4.00 per million tokens, blended at three input tokens per output token. No price change has been announced for either.
- Is GPT-6.1 Sol or Kimi K2.5 better for coding?
- GPT-6.1 Sol scores higher in coding (72 vs 56 on the category index, where 50 is average). On Arena WebDev, the top shared coding benchmark, it is 1752 for GPT-6.1 Sol against 1437 for Kimi K2.5.
- Which is faster, GPT-6.1 Sol or Kimi K2.5?
- Kimi K2.5 streams faster (87 vs 79 output tokens per second, Artificial Analysis 7-day median).
- Is GPT-6.1 Sol or Kimi K2.5 better for agents and tool use?
- GPT-6.1 Sol scores higher in agentic tasks (64 vs 45 on the category index, where 50 is average).
More questions (4)
- Which is better for long documents?
- GPT-6.1 Sol scores higher on long-context benchmarks (66 vs 64). GPT-6.1 Sol accepts more context: 1,050,000 against 262,144 tokens for Kimi K2.5.
- Which has lower latency?
- Kimi K2.5 answers first: 36.0 s against 249.5 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-6.1 Sol vs Kimi K2.5?
- Across the 22 benchmarks both have, GPT-6.1 Sol leads 22, Kimi K2.5 leads 0. Every overall index that rates both puts GPT-6.1 Sol ahead.
- When was GPT-6.1 Sol released?
- GPT-6.1 Sol was released on 29 Sept 2026.