Kimi K2.7 Code vs Kimi K3
At a Glance
| Compare |  Kimi K2.7 CodeMoonshot AI | Kimi K3Moonshot AI |
|---|---|---|
| Intelligence, Cost, and Efficiency | ||
| IntelligenceHigher is better · MM Intelligence v4.4 | #33 of 5727.0 score · 1/3 sources · single-source provisional · missing ARC-AGI-2, LMArena Text · full-core range 9.0–75.7 | #13 of 5773.3 score · 3/3 sources · complete |
| CostLower is better · Published-token output estimate | #8 of 27$0.044 per LiveBench case | #19 of 27$0.163 per LiveBench case |
| EfficiencyHigher is better · MM Efficiency v1.9 | #23 of 2737.8 score · 1/3 sources · single-source provisional · missing ARC-AGI-2, LMArena Text · full-core range 28.8–62.2 | #12 of 2750.1 score · 3/3 sources · complete |
| Pricing and Limits | ||
| Input priceFrom · USD / 1M tokens | $0.6712Openrouter ↗ · Oct 20, 2026 | $0.239Openrouter ↗ · Oct 20, 2026 |
| Output priceFrom · USD / 1M tokens | $3.35Openrouter ↗ · Oct 20, 2026 | $11.936Openrouter ↗ · Oct 20, 2026 |
| Context windowMaximum documented tokens | 262K | 1,049K |
| Model facts checked | Sep 3, 2026View model evidence → | Aug 28, 2026View model evidence → |
Token prices are the lowest available sourced USD rates; input and output may use different providers. Cost ranking estimates output spend on LiveBench, not a full request bill. Ranking methodology →
Available Benchmarks
| Benchmark | Kimi K2.7 Code | Kimi-K3 |
|---|---|---|
| LiveBench2026-06-25-90f8d44da965 · overall · leader | 71.4588% of row best · percent · kimi-k2.7-code · 13,007 output tokens / case | 81.02100% of row best · percent · kimi-k3 · 13,647 output tokens / case |
| Overall ResultCounted from the protocol-matched rows above | 0 benchmark winsNo overall winner | 1 benchmark winNo overall winner |
Third-party benchmark Only like-for-like primary-publisher results are shown; raw scores, relative scores, configuration, and token spend remain visible.
Side-by-Side Facts
| Field | Kimi K2.7 Code | Kimi-K3 |
|---|---|---|
| Developer | Moonshot AI | Moonshot AI |
| Family | Kimi K2 7 | Kimi K3 |
| Model | Kimi K2.7 Code | Kimi-K3 |
| Version | Kimi K2.7 Code | Kimi-K3 |
| Lifecycle | active | active |
| Released | 2026-06-11 | 2026-07-16 |
| Knowledge cutoff | Unknown | Unknown |
| Input modalities | Text, Image, Video | Text, Image |
| Output modalities | Text | Text |
| Context window | 262K | 1,049K |
| Total parameters | 1T | 2.78T |
| Active parameters | 32B | 104B |
| License | modified-mit | other |
| Open weights | Yes | Yes |
| API available | Yes | Yes |
| Self-hostable | Yes | Yes |
| Provider access | Deepinfra (Standard), Fireworks Ai (Standard), Openrouter (Standard), Together Ai (Standard) | Deepinfra (Standard), Fireworks Ai (Standard), Hugging Face (Standard), Openrouter (Standard), Together Ai (Standard) |
| Capabilities | agents, chat, coding, reasoning, tools, vision | chat, generation, reasoning |
Kimi K2.7 Code Capabilities
Kimi K3 Capabilities
Related Comparisons
All image comparisons →| A | Pair | B | Context |
|---|---|---|---|
Kimi-K2.6Moonshot AI | vs | Kimi K2.7 CodeMoonshot AI | family variantsimage, text |
Kimi K2.7 CodeMoonshot AI | vs |  Qwen3.8-MaxQwen | cross-developer peersimage, text, video |
 DeepSeek-V4-Pro-0813DeepSeek | vs | Kimi K2.7 CodeMoonshot AI | cross-developer peerstext |
 Muse Spark 1.3Meta | vs | Kimi K2.7 CodeMoonshot AI | cross-developer peersimage, text, video |
Kimi-K2-ThinkingMoonshot AI | vs | Kimi K2.7 CodeMoonshot AI | family variantstext |
Kimi-K2.5Moonshot AI | vs | Kimi K2.7 CodeMoonshot AI | family variantsimage, text |
Kimi K2.7 CodeMoonshot AI | vs | Qwen3 Coder PlusQwen | cross-developer peerstext |
Kimi K2.7 CodeMoonshot AI | vs | Qwen3 VL FlashQwen | cross-developer peersimage, text, video |
Kimi K2.7 CodeMoonshot AI | vs | Qwen3 VL PlusQwen | cross-developer peersimage, text, video |
Kimi K2.7 CodeMoonshot AI | vs | Qwen3.7 MaxQwen | cross-developer peersimage, text, video |
Kimi K2.7 CodeMoonshot AI | vs | Qwen3.7-PlusQwen | cross-developer peersimage, text, video |
Primary Evidence
Sources and Freshness
Questions
Kimi K2.7 Code vs Kimi K3 FAQs
Is Kimi K2.7 Code or Kimi K3 better for coding?+
This comparison does not currently contain a protocol-matched coding benchmark for both Kimi K2.7 Code and Kimi K3, so Model Markets cannot name a coding leader from pricing, context size, or capability labels alone.
Which is cheaper, Kimi K2.7 Code or Kimi K3?+
Kimi K2.7 Code is $0.6712 and Kimi K3 is $0.239 per million tokens, so Kimi K3 is cheaper on this metric. Kimi K2.7 Code is $3.35 and Kimi K3 is $11.936 per million tokens, so Kimi K2.7 Code is cheaper on this metric.
Which has a larger context window, Kimi K2.7 Code or Kimi K3?+
Kimi K3 has the larger sourced context window. Kimi K2.7 Code supports 262,144 tokens and Kimi K3 supports 1,048,576 tokens.
Which performs better in benchmarks, Kimi K2.7 Code or Kimi K3?+
There is no overall benchmark winner: At least two independently verified, protocol-matched benchmarks are required for an overall winner.
Can Kimi K2.7 Code or Kimi K3 be self-hosted?+
Both models have the same sourced self-hosting status: supported. Kimi K2.7 Code is open weight; Kimi K3 is open weight.
Can Kimi K2.7 Code and Kimi K3 understand images?+
Kimi K2.7 Code is documented with image input; Kimi K3 is documented with image input. This reflects supported input modalities, not vision quality.
Which can generate longer answers, Kimi K2.7 Code or Kimi K3?+
Neither model has a sourced maximum output in this comparison, so a larger limit cannot be determined.
Do Kimi K2.7 Code and Kimi K3 support reasoning and tool use?+
Kimi K2.7 Code: reasoning, tool calling, and image input. Kimi K3: reasoning and image input. Feature support does not establish relative quality.
Which is available from more inference providers, Kimi K2.7 Code or Kimi K3?+
Kimi K2.7 Code has 4 sourced provider routes; Kimi K3 has 5, so Kimi K3 has broader tracked availability.
Which offers better value, Kimi K2.7 Code or Kimi K3?+
There is no universal value winner. Compare the input and output prices above with the matched benchmark result for your workload: cheaper tokens can be offset by different quality, token usage, latency, or provider availability.