Compare / Kimi K3.1 Pro vs GPT-6 Sol
Kimi K3.1 Pro vs GPT-6 Sol
Which is cheaper, which is more capable, and which does better at each kind of work, from 64 results on 5 sources that measured both models.
Moonshot AIKimi K3.1 Pro
OpenAIGPT-6 Sol
Kimi K3.1 Pro wins 2117 too close to callGPT-6 Sol wins 26
Of the 64 results both models have, the better value on each. Results whose reported ranges overlap are too close to call.
Results as of 10 November 2026 (catalogue release 2026-11-10-8342a7178929). See also Moonshot AI vs OpenAI, every model from both labs.
Kimi K3.1 Pro or GPT-6 Sol? The short answer
They score about the same, so pick GPT-6 Sol: it costs 5.3 times less per product listing on CatalogBench.
- Spring Prompt overall: Kimi K3.1 Pro 68, GPT-6 Sol 67 (out of 100; how the score works).
- GPT-6 Sol is cheaper to run: $0.009 against $0.045 per product listing on CatalogBench (5.3 times less).
- Kimi K3.1 Pro is newer: released 23 Oct 2026, against 22 Sep 2026 for GPT-6 Sol.
- On our own benchmarks, on the results both models have: Kimi K3.1 Pro does better for marketing planning; GPT-6 Sol for decks from an analysis.
- On third-party benchmarks, on the results both models have: Kimi K3.1 Pro does better for agents and tool use and reasoning and knowledge.
Head to head
68
Spring Prompt overall
67
$3.81
Price per million tokens, blendedlower is better
$4.00
1,048,576
Context window, tokens
1,050,000
Every fact side by side
| Kimi K3.1 Pro | GPT-6 Sol | |
|---|---|---|
| Price per million tokens | $1.75 in · $10.00 out | $2.00 in · $10.00 out |
| Blended (3 in : 1 out) | $3.81 | $4.00 |
| Cheapest host, blended | $2.38 (Relace, FP4) | $4.00 (Amazon Bedrock) |
| Cost per product listing (CatalogBench) | $0.045 | $0.009 |
| Cached input, per million | – | $0.20 cached input |
| Intelligence Index | – | 47.6 |
| Spring Prompt overall | 68 | 67 |
| Output speed, tokens a second | – | – |
| Context window | 1,048,576 tokens | 1,050,000 tokens |
| Released | 23 Oct 2026 (newer) | 22 Sep 2026 |
| Weights | Open | Closed |
| Developer based in | China | the US |
Prices are the developer's list price where we have read it, otherwise the typical host on OpenRouter; see every model's price. Intelligence and speed from Artificial Analysis.
Which is better for what
Each model's rank among the models on sale today, for each kind of work, from our own benchmarks and licensed sources. Further right is better; the better rank is in green.
Product listingsTurning a sparse product feed and photos into listings that can go live
not measured
3 of 29
Decks from an analysisTurning a finished analysis into a deck you could present as it is
12 of 29
2 of 29
User surveysPlanning a user survey and reading its results without being misled
18 of 20
not measured
Marketing planningPlanning a year of ad spend without overspending
4 of 29
5 of 29
Agents and tool useMulti-step tasks with tools: support desks, coding agents, function calls
1 of 159
8 of 159
Professional workReal tasks from banking, consulting and law, business documents and freelance projects
not measured
13 of 51
Reasoning and knowledgeHard questions across science, maths and general knowledge
9 of 187
14 of 187
WritingWhat people prefer in blind comparisons, and judged writing quality
not measured
50 of 169
Sticking to the factsSummarising without inventing things, and factual answers
not measured
48 of 139
Speed under pressureGood decisions against a real clock (fast chess)
24 of 32
11 of 32
Their slides, side by side
The same brief for both: Manchester store investment review for Hearthside Coffee, an invented company. First slide of each deck from DeckBench.
From our research
Every shared result
Each model's best result on every measure both have, grouped by benchmark, with the setting that produced it. Under each measure, every model that benchmark has measured is a grey tick, best to the right, so you can see where the two sit in the field. The headline measures of each benchmark come first; the rest are under "Show all". Green marks the better value; where a source reports ranges that overlap, neither is marked.
| Measure | Kimi K3.1 Pro | GPT-6 Sol |
|---|---|---|
| BulletBench measured by Spring Prompt · 3 results | ||
| Cost per game · Bullet 60sUS dollars, lower is better 28 · 32 of 38 measured | $0.0175low reasoning | $0.0297no reasoning |
| Ladder Elo · Bullet 60sladder Elo, higher is better 31 · 19 of 38 measured | 147low reasoning | 445no reasoning |
| Games lost on time · Bullet 60s% of games, lower is better 34 · 21 of 38 measured | 87.5%low reasoning | 50.0%no reasoning |
| CatalogBench measured by Spring Prompt · 3 results | ||
| Cost per productUS dollars, lower is better 27 · 16 of 30 measured | $0.0454 | $0.0085 |
| Channel compliance% of products, higher is better 1 · 1 of 30 measured | 100.0% | 100.0% |
| Channel rules broken% of products, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| DeckBench measured by Spring Prompt · 3 results | ||
| Cost per deckUS dollars, lower is better 27 · 23 of 30 measured | $0.54 | $0.21 |
| Deck ratingrating, higher is betterThe rating weighs each win by how strong the other deck was; the win rate counts wins against whichever decks a model happened to meet, so the two can disagree. 12 · 2 of 30 measured | 1,238 | 1,470 |
| Accurate decks% of tasks, higher is better 9 · 5 of 30 measured | 66.7% | 83.3% |
| ROASBench measured by Spring Prompt · 3 results | ||
| Overall scorescore out of 100, higher is better 4 · 5 of 30 measured | 55.1 | 54.1 |
| Audience scorescore out of 100, higher is better 8 · 2 of 30 measured | 67.6 | 69.4 |
| Business scorescore out of 100, higher is better 3 · 8 of 30 measured | 58.2 | 52.9 |
| Artificial Analysis reported by Artificial Analysis · 3 results | ||
| Humanity's Last Exam% of questions, higher is better 9 · 16 of 250 measured | 52.5% | 47.9%max reasoning |
| Terminal-Bench 4.0% of tasks, higher is better 1 · 11 of 120 measured | 67.2% | 43.9%max reasoning |
| Long-context reasoning (AA-LCR)% of questions, higher is better 3 · 11 of 241 measured | 86.3% | 83.7%high reasoning |
Show all 64 results (49 more)
| Measure | Kimi K3.1 Pro | GPT-6 Sol |
|---|---|---|
| BulletBench measured by Spring Prompt · 2 results | ||
| Invalid moves · Bullet 60s% of moves, lower is better 1 · 1 of 38 measured | 0.0%low reasoning | 0.0%no reasoning |
| Median move time · Bullet 60smilliseconds, lower is better 38 · 22 of 38 measured | 5.0 slow reasoning | 1.6 sno reasoning |
| CatalogBench measured by Spring Prompt · 26 results | ||
| Channel rules broken · sales brief% of products, lower is better 1 · 16 of 30 measured | 0.0% | 0.6% |
| Claims to check% of products, lower is better 1 · 1 of 28 measured | 0.0% | 0.0% |
| Claims to check · sales brief% of products, lower is better 1 · 1 of 28 measured | 0.0% | 0.0% |
| Conflicts caught% of conflicts, higher is better 1 · 1 of 30 measured | 100.0% | 100.0% |
| Content quality% of checks, higher is better 1 · 8 of 30 measured | 98.9% | 97.8% |
| Content quality · sales brief% of checks, higher is better 24 · 7 of 30 measured | 95.3% | 98.3% |
| Decision accuracy% of decisions, higher is better 1 · 7 of 30 measured | 98.9% | 98.0% |
| Failed outputs% of products, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| Failed outputs · sales brief% of products, lower is better 28 · 1 of 30 measured | 3.6% | 0.0% |
| Field accuracy% of missing fields, higher is better 3 · 14 of 30 measured | 95.6% | 93.7% |
| Invented values% of filled values, lower is better 21 · 10 of 30 measured | 2.1% | 1.0% |
| Missing UK information% of products, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| Missing UK information · sales brief% of products, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| Not findable% of products, lower is better 17 · 22 of 30 measured | 5.4% | 7.7% |
| Not findable · sales brief% of products, lower is better 14 · 24 of 30 measured | 3.6% | 7.1% |
| Publish-ready listings% of products, higher is better 2 · 5 of 30 measured | 78.6% | 75.0% |
| Publish-ready listings · sales brief% of products, higher is better 2 · 1 of 30 measured | 75.0% | 76.2% |
| Time per productseconds, lower is better 30 · 13 of 30 measured | 114 | 10.7 |
| Unsupported claims% of products, lower is better 1 · 5 of 30 measured | 0.0% | 1.8% |
| Unsupported claimsclaims per product, lower is better 1 · 5 of 30 measured | 0.00 | 0.08 |
| Unsupported claims · sales brief% of products, lower is better 1 · 4 of 30 measured | 1.8% | 3.0% |
| Unsupported claims · sales briefclaims per product, lower is better 5 · 3 of 30 measured | 0.21 | 0.09 |
| Wrong attributes% of products, lower is better 13 · 6 of 30 measured | 16.1% | 14.3% |
| Wrong attributes · sales brief% of products, lower is better 5 · 4 of 30 measured | 14.3% | 13.7% |
| Wrong category or variant% of products, lower is better 5 · 24 of 30 measured | 1.8% | 3.6% |
| Wrong category or variant · sales brief% of products, lower is better 27 · 20 of 30 measured | 5.4% | 3.0% |
| DeckBench measured by Spring Prompt · 14 results | ||
| Caveat dropped% of tasks, lower is better 14 · 1 of 30 measured | 16.7% | 0.0% |
| Clean layout% of tasks, higher is better 27 · 8 of 30 measured | 0.0% | 66.7% |
| Design quality% of the maximum, higher is better 3 · 2 of 30 measured | 83.9% | 87.9% |
| Draft figure quoted% of tasks, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| Findings missing% of tasks, lower is better 28 · 20 of 30 measured | 33.3% | 16.7% |
| Head-to-head win rate% of comparisons, higher is betterThe rating weighs each win by how strong the other deck was; the win rate counts wins against whichever decks a model happened to meet, so the two can disagree. 10 · 2 of 30 measured | 71.1% | 88.1% |
| Layout defects% of tasks, lower is better 27 · 8 of 30 measured | 100.0% | 33.3% |
| Misleading metric used% of tasks, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| Presentable decks% of tasks, higher is better 14 · 3 of 30 measured | 0.0% | 50.0% |
| Recommendation late or wrong% of tasks, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| Slides needing work% of tasks, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| Time per deckseconds, lower is better 30 · 20 of 30 measured | 1,288 | 169 |
| Unsupported claims% of tasks, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| Unsupported numbers% of tasks, lower is better 1 · 1 of 30 measured | 0.0% | 0.0% |
| ROASBench measured by Spring Prompt · 6 results | ||
| Consistency scorescore out of 100, higher is better 6 · 3 of 30 measured | 36.3 | 42.9 |
| Contribution profitsimulated US dollars, higher is better 4 · 14 of 30 measured | $858,023 | $600,357 |
| Cost of a runUS dollars, lower is better 28 · 19 of 30 measured | $1.60 | $0.24 |
| Months over budgetmonths of 12, lower is better 1 · 1 of 30 measured | 0 | 0 |
| Planning scorescore out of 100, higher is better 4 · 14 of 30 measured | 55.0 | 54.8 |
| Return on ad spendprofit per $1 spent, higher is better 2 · 9 of 30 measured | $0.72 | $0.65 |
| Artificial Analysis reported by Artificial Analysis · 1 result | ||
| SciCode% of problems, higher is better 7 · 18 of 121 measured | 60.8% | 57.6%max reasoning |
Compare them with other models
- Moonshot AI vs OpenAIEvery model from both labs, best against best
- Kimi K3.1 Pro vs Claude Opus 5.5Highest-scoring overall at $3–10 per million tokens
- Kimi K3.1 Pro vs Claude Fable 5.2Highest-scoring overall available today
- Kimi K3.1 Pro vs Muse WonderJust above overall
- GPT-6 Sol vs Claude Opus 5.5Most intelligent at $3–10 per million tokens
- GPT-6 Sol vs DeepSeek-V4.1-Flash-1017Cheapest with similar intelligence
- GPT-6 Sol vs Claude Fable 5.2Highest-scoring overall available today
Quick answers
Is Kimi K3.1 Pro better than GPT-6 Sol?
They score about the same, so pick GPT-6 Sol: it costs 5.3 times less per product listing on CatalogBench.
Which is cheaper, Kimi K3.1 Pro or GPT-6 Sol?
GPT-6 Sol is cheaper to run: $0.009 against $0.045 per product listing on CatalogBench (5.3 times less).
Which is better for decks from an analysis, Kimi K3.1 Pro or GPT-6 Sol?
GPT-6 Sol ranks 2 of 29 models on sale for turning a finished analysis into a deck you could present as it is, against 12 for Kimi K3.1 Pro.
Which is better for marketing planning, Kimi K3.1 Pro or GPT-6 Sol?
Kimi K3.1 Pro ranks 4 of 29 models on sale for planning a year of ad spend without overspending, against 5 for GPT-6 Sol.
Which is better for agents and tool use, Kimi K3.1 Pro or GPT-6 Sol?
Kimi K3.1 Pro ranks 1 of 159 models on sale for multi-step tasks with tools: support desks, coding agents, function calls, against 8 for GPT-6 Sol.
Which is better for reasoning and knowledge, Kimi K3.1 Pro or GPT-6 Sol?
Kimi K3.1 Pro ranks 9 of 187 models on sale for hard questions across science, maths and general knowledge, against 14 for GPT-6 Sol.
Which is better for speed under pressure, Kimi K3.1 Pro or GPT-6 Sol?
GPT-6 Sol ranks 11 of 32 models on sale for good decisions against a real clock (fast chess), against 24 for Kimi K3.1 Pro.
Which has the bigger context window, Kimi K3.1 Pro or GPT-6 Sol?
GPT-6 Sol: 1,050,000 tokens, against 1,048,576 for Kimi K3.1 Pro.
Run Kimi K3.1 Pro and GPT-6 Sol on your own prompt.
Benchmarks aren't your task. Test both side by side on your prompt, with every answer checked against your rules and the cost of each one.