Kimi K2.7 Code
Moonshot AI · released Jun 11, 2026 · moonshotai/Kimi-K2.7-Code
- Type
- Open weightsCustom licence
- Params
- 1.1T
- Context
- 262K
32B active per word · about 197K words of context · download allowed, licence restricts use
Our take
Written Aug 2, 2026Moonshot's very large coding model can be downloaded. At over a trillion parameters it is an unusually large model to be downloadable at all, aimed at agentic coding workloads where scale matters.
Choose this for agentic coding workloads where frontier-scale capability matters and provider portability is useful. Skip it if your licence cannot tolerate custom restricted terms, or if you need to self-host on standard hardware.
The case for it
- Over one trillion total parameters — among the largest models you can actually download.
- More than a dozen hosts serve it at near-identical prices, which keeps the market competitive.
The case against it
- Custom restricted licence; legal review is needed before commercial fine-tuning or redistribution.
- Over a trillion parameters. Even compressed, it is far beyond any single consumer GPU or Mac.
How good is it?
IntelligencePuzzles, maths, exam questions
Kimi K2.7 Code is not on Arena Text (overall), which is where the rating would come from, so there is no rating here. It is on LiveBench Reasoning, in 19th of 35 with 82.8.
CodingWriting and fixing code on its own
Kimi K2.7 Code is not on Arena Coding, which is where the rating would come from, so there is no rating here. It is on Arena Code (WebDev), in 26th of 74 with 1472.6.
AgenticPlanning, calling tools, staying on task
Arena Agent (IPS)17th of 36 · 0.004
WritingWe do not rate this
Two boards come close and neither tests writing: Arena Creative Writing asks people which of two replies they prefer, and LiveBench Language tests whether a model understood a passage. So we show where Kimi K2.7 Code placed and give it no mark out of five.
These tests check whether a model follows instructions — a precondition for all the work above, but not a measure of how well that work is done, which is why they get no rating.
Every published score for this model10 scoresEvery figure we hold, from 10 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Can you run it yourself?
- Fits in memory
- weights load entirely on the card
- Spills to system RAM
- some weights offload; much slower
- Too large
- will not load even with offload
- est
- size is calculated; the verdict could change by 10%
GeForce RTX 4090 · 24 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Apple M1 Pro (16-core GPU) · 32 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Radeon RX 7900 XT · 20 GB
Too large for this card. The weights do not fit even with part of them offloaded to system memory.
Memory use by level
Against a 24 GB card.
All 3 sizes here are calculated, not measured. We hold no measured file for this model, so each size comes from the parameter count and every verdict above inherits that uncertainty.
Check against your own machine → · All 71 devices, with every size →
Or rent it from someone else
Cheapest of 20 live listings. Picked at the widest standard context we hold, within one quantisation slice, so the numbers beside it are a price one host actually charges.
- per 1M tokens
- $0.73 in / $3.50 out
- Context served
- 262K
- Throughput
- Not measured
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| OpenRouter | $0.73 / $3.50 | 262K | not measured | Unknown | Unknown | Unknown |
| DeepInfrafp4 | $0.74 / $3.50 | 262K | not measured | Unknown | Unknown | Unknown |
| DeepInfrafp4 | $0.74 / $3.50 | 262K | 80 tok/s | No | No | Confirmed |
| Venice AIint4 | $0.75 / $3.50 | 256K | 74 tok/s | No | No | Confirmed |
| Parasailint4 | $0.76 / $3.50 | 262K | 75 tok/s | No | No | Confirmed |
| CoreWeaveint4 | $0.71 / $3.50 | 262K | 176 tok/s | No | No | Confirmed |
| Inceptronint4 | $0.72 / $3.50 | 262K | 128 tok/s | No | No | Confirmed |
| Ambientint4 | $0.73 / $3.50 | 262K | 42 tok/s | No | Yesunknown period | Unknown |
| ModelRunfp4 | $0.85 / $3.75 | 262K | 145 tok/s | No | No | Confirmed |
| SiliconFlowfp8 | $0.86 / $3.80 | 262K | 71 tok/s | No | No | Confirmed |
| Novita AIint4 | $0.91 / $3.84 | 262K | 3 tok/s | No | No | Confirmed |
| Fireworks AI | $0.95 / $4.00 | 262K | 44 tok/s | No | No | Confirmed |
| Moonshot AIint4 | $0.95 / $4.00 | 262K | 3 tok/s | No | No | Confirmed |
| Novita AI | $0.95 / $4.00 | 262K | not measured | Unknown | Unknown | Unknown |
| Together AI | $0.95 / $4.00 | 262K | 166 tok/s | No | No | Confirmed |
| Alibaba Cloudfp8 | $0.95 / $4.00 | 262K | 20 tok/s | No | Yesunknown period | Unknown |
| Cloudflare Workers AI | $0.95 / $4.00 | 262K | 47 tok/s | No | Yesunknown period | Unknown |
| AtlasCloudint4 | $0.95 / $4.00 | 262K | 43 tok/s | No | Yesunknown period | Unknown |
| Nebius AI Studiofp4 | $0.95 / $4.00 | 8K | 177 tok/s | No | No | Unknown |
| Moonshot AIhighspeed tierint4 | $1.90 / $8.00 | 262K | 78 tok/s | No | No | Unknown |
DeepInfra
Novita AI
Fireworks AI
Together AI
Cloudflare Workers AI
Nebius AI StudioAcross the 20 listings we hold: 17 say they do not train on prompts, 0 say they do and 3 do not say. 11 appear in the zero-retention registry we check; the rest are unknown to us rather than confirmed either way.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
- ✓
- Supported
- ✗
- Not supported
- Not published
- host gave no parameter list
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| OpenRouter | ✓ | ✓ | ✓ |
| DeepInfrafp4 | |||
| DeepInfrafp4 | ✓ | ✓ | ✗ |
| Venice AIint4 | ✓ | ✓ | ✓ |
| Parasailint4 | ✓ | ✓ | ✓ |
| CoreWeaveint4 | ✓ | ✓ | ✓ |
| Inceptronint4 | ✓ | ✓ | ✓ |
| Ambientint4 | ✓ | ✓ | ✓ |
| ModelRunfp4 | ✓ | ✓ | ✓ |
| SiliconFlowfp8 | ✓ | ✗ | ✗ |
| Novita AIint4 | ✓ | ✓ | ✓ |
| Fireworks AI | ✓ | ✓ | ✓ |
| Moonshot AIint4 | ✓ | ✓ | ✓ |
| Novita AI | |||
| Together AI | ✓ | ✓ | ✓ |
| Alibaba Cloudfp8 | ✓ | ✓ | ✓ |
| Cloudflare Workers AI | ✓ | ✓ | ✓ |
| AtlasCloudint4 | ✓ | ✓ | ✓ |
| Nebius AI Studiofp4 | ✓ | ✓ | ✓ |
| Moonshot AIhighspeed · int4 | ✓ | ✓ | ✓ |
Tool calling: 18 of 20 listings say yes, 2 publish no parameter list. JSON output: 17 of 20 listings say yes, 1 says no, 2 publish no parameter list. Strict schema: 16 of 20 listings say yes, 2 say no, 2 publish no parameter list.
Models people weigh against Kimi K2.7 Code
When we formed this view
Dates behind this page
Prices last checked 3d ago
What we do not know about this model yet
- We hold no measured file for it, so all 3 sizes on this page are calculated from the parameter count.
- 2 of 20 listings publish no parameter list, so what their API accepts is unknown to us.
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 3 of 20 listings do not say whether they train on prompts.
Licence and identifiers
What the licence allowsCustom licence, what it allows commercially, and the identifiers you need to pull this model — its Hugging Face repo, our slug and a machine-readable card.
Licence
Custom licence
This model ships custom license terms that don't map to a known template. We haven't parsed them, so commercial use, redistribution and derivatives are unverified — review the original terms before shipping.
Identifiers
- Hugging Face
- moonshotai/Kimi-K2.7-Code
- Architecture
- Mixture of experts
- Modality record
- text+image->text
- Catalogue slug
- moonshotai-kimi-k2-7-code