text model comparison
Claude Fable 5.1 vs Kimi K3.1 Pro
Claude Fable 5.1 and Kimi K3.1 Pro are both current text models, from Anthropic and Moonshot Kimi. Kimi K3.1 Pro is 80% cheaper per 1M input tokens than Claude Fable 5.1, $2 vs $10. On Arena WebDev, Claude Fable 5.1 scores 1746 Elo and Kimi K3.1 Pro 1695 Elo (Arena, measured October 28, 2026). Both accept about 1.05M tokens of context (1,048,576 and 1,000,000).
- Higher Arena WebDev
- Claude Fable 5.11746 Elo against 1695 Elo
- Lower price per 1M input tokens
- Kimi K3.1 Pro$2 against $10
- Longer context window
- Kimi K3.1 Pro1.05M tokens against 1M tokens
- Newer release
- Kimi K3.1 ProOct 2026 against Sep 2026
Data verified November 5, 2026 · Dataset 2026-11-05
Side by side
Claude Fable 5.1 and Kimi K3.1 Pro, line by line
Prices in effect today, compared at the same tier wherever both list one. The better value in each row is highlighted.
| Fact | Claude Fable 5.1 | Kimi K3.1 Pro |
|---|---|---|
| Price | ||
| Price per 1M input tokens | $10 | $2 (better) |
| Price per 1M output tokens | $50 | $10 (better) |
| Independent benchmarks | ||
| Arena WebDev | 1746 Elo (#6 of 46) (better) | 1695 Elo (#7 of 46) |
| Arena Text | 1504 Elo (#5 of 52) | Not measured |
| FrontierMath Tiers 1-3 (v2) | 90.2% (#5 of 41) | Not measured |
| SimpleQA Verified | 70.8% (#6 of 38) | Not measured |
| Specs | ||
| Provider | Anthropic | Moonshot Kimi |
| Release date | September 1, 2026 | October 23, 2026 |
| Access | API only | API and open weights |
| Context window | 1,000,000 tokens | 1,048,576 tokens |
| Max output | 128,000 tokens | 1,048,576 tokens |
| Input | text, image | text, image, video |
| Output | text | text |
Provider results
What the labs report
Results tables a lab published with one of its models, where it lists both Claude Fable 5.1 and Kimi K3.1 Pro. These are the lab's own figures, not independent measurements, and each table is read on its own: labs run benchmarks with different harnesses and budgets.
Moonshot AI, Kimi K3.1 Pro model card, November 2, 2026Moonshot AI · 2026-11-05
| Benchmark | Claude Fable 5.1 | Kimi K3.1 Pro |
|---|---|---|
| Terminal-Bench 4.0 | 57.9% | 65.7% (better) |
| Terminal-Bench-Science 0.1 | 40.0% (better) | 38.1% |
| SWE Marathon v1.1 | 45.6% | 51.7% (better) |
| DeepSWE v1.1 | 67.4% | 75.5% (better) |
| ProgramBenchtasks at least 95% solved | 53.5% (better) | 31.5% |
| PostTrainBench v1.2 | 44.6% (better) | 36.5% |
| BenchCADwithout tools | 60.6% | 72.8% (better) |
| BenchCADwith tools | 92.6% | 92.7% (better) |
| HLE-Fullwith tools | 65.6% | 70.0% (better) |
| HLE-Fullwithout tools | 57.6% (better) | 51.1% |
| HLE-Diamond | 55.1% (better) | 39.4% |
| GPQA-Diamond | 93.7% | 94.2% (better) |
| BioMysteryBench | 81.6% (better) | 80.6% |
| APEX-Agents | 68.6% | 69.4% (better) |
| OSWorld 2.1 | 80.7% (better) | 79.6% |
| OfficeQA Pro v2 | 54.5% | 64.4% (better) |
| Frontier Finance | 60.8% (better) | 58.6% |
| HANDBOOK.md | 38.8% | 44.6% (better) |
| Harvey Legal Agent | 19.1% | 23.7% (better) |
| GDP.pdf | 27.6% | 34.6% (better) |
| GDP.pdfwith Python | 83.8% | 85.7% (better) |
| MMMU-Pro | 86.3% (better) | 84.1% |
| MMMU-Prowith Python | 88.4% | 90.0% (better) |
| CharXiv (RQ) | 88.8% | 89.0% (better) |
| CharXiv (RQ)with Python | 92.7% | 93.4% (better) |
| ChartoGraphywith Python | 88.4% | 90.2% (better) |
| BabyVisionwith Python | 95.1% | 95.4% (better) |
| ZeroBench mainpass@5 | 35.0% | 39.0% (better) |
| ZeroBench mainwith Python, pass@5 | 59.0% | 63.0% (better) |
| PerceptionBench | 63.0% | 68.3% (better) |
| ExploitBench | 83.0% (better) | 74.7% |
| ExploitGym | 30.4% (better) | 20.7% |
Moonshot's table from the Kimi K3.1 Pro model card, every model at its highest reasoning effort; rivals' scores are the highest their labs or the benchmarks' leaderboards report, unless Moonshot ran them (its footnotes say which). DeepSeek-V4.1-Flash is the 1017 build. Left out here: Moonshot's internal benchmarks, the rows Artificial Analysis runs and the hallucination rows, where lower is better.
Charts
Against every current text model
Where Claude Fable 5.1 and Kimi K3.1 Pro land among the current models of their task, on every independent benchmark, on price and on context.
Where each stands
On Arena WebDev, Claude Fable 5.1 ranks #6 and Kimi K3.1 Pro #7 of 46 current models. Each bar is the share of current models it beats: on score, price and context.
Independent benchmarks
Arena Text
Claude Fable 5.1:#5of 52
1,504 Elo · ahead of 90%
Kimi K3.1 Pro: Not scored
Arena WebDev
Claude Fable 5.1:#6of 46
1,746 Elo · ahead of 89%
Kimi K3.1 Pro:#7of 46
1,695 Elo · ahead of 87%
FrontierMath Tiers 1-3 (v2)
Claude Fable 5.1:#5of 41
90.2% · ahead of 90%
Kimi K3.1 Pro: Not scored
SimpleQA Verified
Claude Fable 5.1:#6of 38
70.8% · ahead of 86%
Kimi K3.1 Pro: Not scored
Price and context
Price per 1M input tokens
Claude Fable 5.1:#51of 55
$10 · cheaper than 2%
Kimi K3.1 Pro:#30of 55
$2 · cheaper than 24%
Price per 1M output tokens
Claude Fable 5.1:#51of 55
$50 · cheaper than 2%
Kimi K3.1 Pro:#35of 55
$10 · cheaper than 26%
Context window
Claude Fable 5.1:#26of 64
1M tokens · longer than 22%
Kimi K3.1 Pro:#11of 64
1.05M tokens · longer than 67%
Quality against price
Claude Fable 5.1 scores 1,504 at $20 (blended price per 1M tokens). The line joins the current models no other beats on both price and Arena Text score; hover any point to name it.
Quality against price: the data
| Model | Provider | Blended price per 1M tokens | Arena Text (Elo) |
|---|---|---|---|
| Gemini 4 Carbon | $4 | 1,528 | |
| Gemini 4 Argon | $4 | 1,520 | |
| Claude Opus 5.5 | Anthropic | $8 | 1,509 |
| Muse Wonder | Meta | $8 | 1,507 |
| Claude Fable 5 | Anthropic | $20 | 1,504 |
| Claude Fable 5.1 | Anthropic | $20 | 1,504 |
| Gemini 3.8 Flash | $1.50 | 1,496 | |
| Muse Spark 1.3 | Meta | $2 | 1,495 |
| Claude Opus 5 | Anthropic | $10 | 1,492 |
| Kimi K3 | Moonshot Kimi | $6 | 1,488 |
| GPT-5.6 Sol | OpenAI | $8 | 1,485 |
| Gemini 3.6 Flash | $3 | 1,483 | |
| GPT-5.5 | OpenAI | $11 | 1,482 |
| Claude Opus 4.8 | Anthropic | $10 | 1,482 |
| Qwen3.8-Max | Alibaba Qwen | $3 | 1,482 |
| GPT-6.1 Sol | OpenAI | $4 | 1,481 |
| GLM-5.3 | Z.ai | $2.15 | 1,481 |
| MiMo-V2.6-Pro | Xiaomi MiMo | $0.54 | 1,480 |
| Claude Sonnet 5.5 | Anthropic | $4 | 1,477 |
| GLM-5.3-Flash | Z.ai | $0.24 | 1,477 |
| GPT-6 Astra | OpenAI | $20 | 1,476 |
| DeepSeek-V4.1-Flash | DeepSeek | $0.52 | 1,476 |
| GPT-5.4 | OpenAI | $5.63 | 1,475 |
| Gemini 3 Flash Preview | $1.13 | 1,473 | |
| Grok 4.20 | xAI | $1.56 | 1,472 |
| GPT-5.6 Terra | OpenAI | $4.50 | 1,467 |
| Grok 4.5 | xAI | $3 | 1,465 |
| DeepSeek-V4-Pro | DeepSeek | $1.98 | 1,465 |
| Claude Sonnet 5 | Anthropic | $4 | 1,462 |
| GPT-6 Sol | OpenAI | $4 | 1,458 |
| Gemini 3.5 Flash-Lite | $0.85 | 1,456 | |
| Grok 4.6 | xAI | $3 | 1,455 |
| GPT-5.6 Luna | OpenAI | $0.45 | 1,453 |
| MiMo-V2.6-Flash | Xiaomi MiMo | $0.18 | 1,448 |
| Step 5 Preview | StepFun | $1.43 | 1,448 |
| Grok 4.3 | xAI | $1.56 | 1,442 |
| GPT-6 Luna | OpenAI | $0.20 | 1,441 |
| Grok 4.7 | xAI | $3 | 1,441 |
| Mistral Large 4 | Mistral AI | $1.05 | 1,432 |
| Claude Haiku 5.5 | Anthropic | $0.20 | 1,425 |
| GPT-4.1 | OpenAI | $3.50 | 1,415 |
| Solar Pro 4 | Upstage | $0.52 | 1,388 |
| GPT-4.1 mini | OpenAI | $0.70 | 1,383 |
| GPT-4.1 nano | OpenAI | $0.18 | 1,322 |
Cost
What each costs
Example workloads at each provider's listed rates, then every other rate either one lists. The cheaper figure in each row is highlighted.
The same work, priced
For 10M input and 2M output tokens, Claude Fable 5.1 costs 5.0 times as much as Kimi K3.1 Pro, at each provider's listed rates.
10M input and 2M output tokens
Claude Fable 5.1 costs 5.0× as much
- Claude Fable 5.1
- $200
- Kimi K3.1 Pro
- $40
The same, with 9M of the input tokens cached
Claude Fable 5.1 costs 4.7× as much
- Claude Fable 5.1
- $112.25
- Kimi K3.1 Pro
- $23.80
| Workload or rate | Claude Fable 5.1 | Kimi K3.1 Pro |
|---|---|---|
| Example workloads | ||
| 10M input and 2M output tokens | $200 | $40 (better) |
| The same, with 9M of the input tokens cached | $112.25 | $23.80 (better) |
| The same, through the batch API | $100 | Not listed |
| Other listed rates | ||
| 1M input tokens, cached | $0.25 | $0.20 (better) |
| 1M input tokens, 5-minute cache write | $12.50 | $2 (better) |
| 1M input tokens, 1-hour cache write | $20 | $4 (better) |
| 1M input tokens, batch | $5 | Not listed |
| 1M output tokens, batch | $25 | Not listed |
| 1M input tokens, Highspeed endpoint | Not listed | $4 |
| 1M output tokens, Highspeed endpoint | Not listed | $20 |
Your own bill will differ. These figures use list prices and example workloads. The AI Model Cost Audit reviews your real usage, shortlists models that score as well for each task for less, and estimates the savings.
Launch price $100, regular price $150
Verdict
When to choose each
Claude Fable 5.1
Where it leads
- Higher Arena WebDev score, 1746 Elo vs 1695 Elo
- Batch pricing, $5 input and $25 output per 1M tokens
Best for
- Long-horizon agentic coding where a large prompt prefix is replayed across many turns
- Work where Claude Opus 5 at higher effort has already been evaluated and fell short
Watch out for
- Base input and output pricing is unchanged from Fable 5 at $10 / $50; the saving is entirely in cache reads, so a workload with little cache reuse sees no discount
- The 4.7-generation tokenizer produces roughly 30% more tokens for the same text than Claude Sonnet 4.6 and earlier, so per-token rates are not directly comparable across generations
- Anthropic’s launch comparisons are self-reported and were not reproduced independently here
Kimi K3.1 Pro
Where it leads
- Lower input and output prices, $2 and $10 vs $10 and $50 per 1M tokens
- Longer responses, up to 1,048,576 vs 128,000 output tokens
- Accepts video input
- Open weights you can run yourself
- Cheaper cached input, $0.20 vs $0.25 per 1M tokens
Best for
- Long-horizon coding and agent work at mid-tier prices
- Long documents, images and video in one 1M-token context
Watch out for
- Thinking is always on: reasoning effort runs low, high or max, and max is the default
- The Highspeed endpoint, two to three times faster by Moonshot's account, costs twice as much at $4 input and $20 output per 1M tokens
- The weights carry Moonshot's own license: a model-as-a-service business above $20M in yearly revenue needs a separate agreement, and products above 100M monthly users or $20M monthly revenue must display the Kimi K3.1 Pro name
- Self-hosting takes a multi-node GPU cluster: the MXFP4 weights alone are about 1.56 TB
- Epoch AI and Arena Text have not scored it yet
From benchmark to production
Building on Claude Fable 5.1 or Kimi K3.1 Pro?
Public scores rarely hold on your own data. We test both models on your real tasks, then build the agent or pipeline around the one that wins.
AI Model Cost AuditLaunch price $100, regular price $150What the audit includes
FAQ
Claude Fable 5.1 vs Kimi K3.1 Pro FAQ
Which is cheaper, Claude Fable 5.1 or Kimi K3.1 Pro?
Kimi K3.1 Pro is 80% cheaper per 1M input tokens than Claude Fable 5.1, $2 vs $10. For 10M input and 2M output tokens, Claude Fable 5.1 costs $200 and Kimi K3.1 Pro $40 at standard rates.
Which is cheaper with prompt caching, Claude Fable 5.1 or Kimi K3.1 Pro?
Claude Fable 5.1 charges $0.25 per 1M cached input tokens, 98% below its standard input rate, and Kimi K3.1 Pro $0.20, 90% below. With 9M of 10M input tokens cached and 2M output tokens, Claude Fable 5.1 costs $112.25 and Kimi K3.1 Pro $23.80.
Which scores higher on independent benchmarks, Claude Fable 5.1 or Kimi K3.1 Pro?
Claude Fable 5.1 leads on Arena WebDev, 1746 Elo to 1695 Elo. Scores are independent evaluations, not either vendor's launch figures.
How do Claude Fable 5.1 and Kimi K3.1 Pro compare in Moonshot AI's own results?
In Kimi K3.1 Pro model card (November 2, 2026), Kimi K3.1 Pro leads on Terminal-Bench 4.0, 65.7% to 57.9%; Claude Fable 5.1 leads on Terminal-Bench-Science 0.1, 40.0% to 38.1%; Kimi K3.1 Pro leads on SWE Marathon v1.1, 51.7% to 45.6%. These are Moonshot AI's figures, not independent measurements.
Which has the larger context window, Claude Fable 5.1 or Kimi K3.1 Pro?
Both accept about 1.05M tokens of context (1,048,576 and 1,000,000). Kimi K3.1 Pro returns up to 1,048,576 tokens per response, Claude Fable 5.1 128,000.
How do Claude Fable 5.1 and Kimi K3.1 Pro differ in features?
Claude Fable 5.1 takes text and image input; Kimi K3.1 Pro takes text, image and video.
Can I run Claude Fable 5.1 or Kimi K3.1 Pro myself?
Kimi K3.1 Pro is released as open weights you can run yourself; Claude Fable 5.1 is available through its provider's API only.
Further reading
More on Claude Fable 5.1 and Kimi K3.1 Pro
Articles on Claude Fable 5.1 and Kimi K3.1 Pro
- Kimi K3.1 Pro: Benchmarks, Price and Should You Switch?October 25, 2026
- Grok 4.7 vs Fable 5.1 vs GPT-6 Astra: Real Cost per TaskSeptember 23, 2026
- Jev vs Fable 5.1 vs GPT-6 Astra: Which Model for Which JobSeptember 22, 2026
- GPT-6 Astra vs Fable 5.1: Two Scoreboards, Two WinnersSeptember 22, 2026
- Claude Fable 5.1: Same Price, 75% Cheaper CacheSeptember 2, 2026
- Fable 5.1 vs Opus 5 vs GPT-5.6: Which to UseSeptember 1, 2026
Related comparisons
- Claude Fable 5.1 vs Gemini 4 Argon
- Claude Opus 5.5 vs Kimi K3.1 Pro
- Claude Fable 5.1 vs Claude Sonnet 5.5
- Claude Sonnet 5.5 vs Kimi K3.1 Pro
- Claude Fable 5.1 vs Claude Opus 5.5
- Claude Sonnet 5 vs Kimi K3.1 Pro
- Claude Fable 5.1 vs Grok 4.7
- DeepSeek-V4.1-Flash vs Kimi K3.1 Pro
- Claude Fable 5.1 vs GPT-6 Astra
- Gemini 4 Argon vs Kimi K3.1 Pro
- Claude Fable 5.1 vs Claude Opus 4.8
- GLM-5.3 vs Kimi K3.1 Pro
Production evaluation
Choosing between Claude Fable 5.1 and Kimi K3.1 Pro?
Public benchmarks narrow the shortlist. We run both on your own tasks, data and cost limits, then build the routing and review controls to run the winner in production.
- A head-to-head evaluation on your real workload
- Routing between models by cost and difficulty