Kimi K2.7 Code kimi-k2.7-code
moonshotai · kimi coding flagship agentic
#7 overall
- params
- 1059B (A32B)
- arch
- moe
- context
- 256k
- license
- modified-mit open weights
- released
- Jun 2026
- reasoning
- yes
- downloads/30d
- 223.1k
# architecture
- layers
- 61
- d_model
- 7168
- heads
- 64q / 64kv
- head dim
- 112
- experts
- top-8 of 384 + 1 shared
- vocab
- 163840
- family
- kimi_k2
signal path of one layer, generated from the registry's structured fields — dimension lines quote the model's real numbers; an MoE trace forks at the router, a dense trace runs straight through. Geometry fields come from the repo's config.json.
# why ranked
Overall: #7 — score 90.0 ● high — all signals present
| signal | weight | input (0–100) |
|---|---|---|
| aa_intelligence | 0.70 | 89.7 |
| bench_composite | 0.30 | 90.7 |
benchmark panel evidence:
| complete panel | score (0–100) |
|---|---|
| livebench-v1 | 90.7 |
Coding: provisional — score 84.2 ● low — a single signal; treat with caution
provisional: too few independent signals for a numbered position — the score is shown, but this model sorts after every ranked model.
| signal | weight | input (0–100) |
|---|---|---|
| aa_coding | 0.30 | 84.2 |
| aider_polyglot | 0.30 | — |
| swe_bench_verified | 0.40 | — |
missing signals are dropped and the remaining weights renormalized — never imputed.
# trend
methodology changed during this history window; score movement across that boundary is not model movement. See methodology v5.
# benchmarks
| benchmark | score | source | date |
|---|---|---|---|
| AA Coding Index | 60.8 | Artificial Analysis | — |
| AA Intelligence Index | 33.9 | Artificial Analysis | — |
| LMArena Elo (Code) | 1472.0 | LMArena | Sep 2026 |
| LiveBench Agentic Coding | 45.7 | LiveBench | Jun 2026 |
| LiveBench Coding | 74.0 | LiveBench | Jun 2026 |
| LiveBench Data Analysis | 62.7 | LiveBench | Jun 2026 |
| LiveBench IF | 56.3 | LiveBench | Jun 2026 |
| LiveBench Language | 77.9 | LiveBench | Jun 2026 |
| LiveBench Mathematics | 79.6 | LiveBench | Jun 2026 |
| LiveBench Reasoning | 82.8 | LiveBench | Jun 2026 |
# where to run
| provider | quant | ctx | $/M in | $/M out | $/M cache | price src | tps | uptime | ✓ |
|---|---|---|---|---|---|---|---|---|---|
| Featherless | — | — | — | — | via hfrouter | — | — | ||
| Nebius | — | — | — | — | nebius | — | — | ✓ | |
| StreamLake | unknown | 250k | $0.71 | $3.00 | $0.14 | via openrouter | — | 99.9% | |
| Inceptron | int4 | 256k | $0.66 | $3.40 | $0.18 | via openrouter | — | 99.8% | |
| DeepInfra | fp4 | 256k | $0.68 | $3.40 | — | deepinfra | — | 99.8% | ✓ |
| CoreWeave | int4 | 256k | $0.71 | $3.50 | $0.15 | via openrouter | — | 100.0% | |
| Wandb | 256k | $0.71 | $3.50 | — | via litellm | — | — | ||
| Venice | int4 | 250k | $0.75 | $3.50 | — | venice | — | 97.5% | ✓ |
| ModelRun | fp4 | 256k | $0.85 | $3.75 | $0.16 | via openrouter | — | 100.0% | |
| SiliconFlow | fp8 | 256k | $0.86 | $3.80 | $0.18 | openrouter | — | 100.0% | ✓ |
| Alibaba | fp8 | 256k | $0.95 | $4.00 | $0.19 | via openrouter | — | 100.0% | |
| AtlasCloud | int4 | 256k | $0.95 | $4.00 | — | atlascloud | — | — | ✓ |
| Azure | 256k | $0.95 | $4.00 | $0.19 | via modelsdev | — | — | ||
| BaseTen | fp4 | 262k | $0.95 | $4.00 | $0.16 | via openrouter | — | 100.0% | |
| Cloudflare | unknown | 256k | $0.95 | $4.00 | $0.19 | via openrouter | — | 99.7% | |
| Fireworks | unknown | 256k | $0.95 | $4.00 | $0.19 | via openrouter | — | 97.5% | |
| GMICloud | fp8 | 256k | $0.95 | $4.00 | $0.19 | via openrouter | — | 97.9% | |
| Moonshot AI | int4 | 256k | $0.95 | $4.00 | $0.19 | via openrouter | — | 100.0% | |
| Novita | int4 | 256k | $0.95 | $4.00 | — | novita | — | 100.0% | ✓ |
| Phala | 256k | $0.95 | $4.00 | — | phala | — | — | ✓ | |
| Qwen Ai Platform | 224k | $0.95 | $4.00 | — | via litellm | — | — | ||
| Qwencloud | 224k | $0.95 | $4.00 | — | via litellm | — | — | ||
| Together | unknown | 256k | $0.95 | $4.00 | $0.19 | via openrouter | — | — | |
| Moonshot AI | int4 | 256k | $1.90 | $8.00 | $0.38 | via openrouter | — | 100.0% |
sorted by blended price ((3·input + output) / 4 per 1M) · ✓ = the provider's own catalog confirms the offer · "via …" prices are what the aggregator routing the offer charges, not the provider's own list price
source aliases
- aa
kimi-k2.7-code- openrouter
moonshotai/kimi-k2.7-code