Kimi · 2025-07 · Sparse MoE
Kimi K2 Base
Kimi K2 Base is a multi-head latent attention (MLA) transformer released by Kimi in 2025-07, with 61 layers, hidden size 7168 and a context window of 131,072 tokens.
Layer stack
Attention ×61 · Layers 61
Key facts
| Family | Kimi |
|---|---|
| Released | 2025-07 |
| Params | 1,040.2 B |
| Active | 32.8 B |
| Context | 131,072 tokens |
| Attention | MLA · kv_lora_rank 512 |
| Layers | 61 |
| Hidden | 7,168 |
| Heads | 64 |
| Vocab | 163,840 |
| Position | RoPE |
| Norm | RMSNorm |
| Activation | silu |
| Dtype | fp8 |
| Architecture class | DeepseekV3ForCausalLM |
Architecture overview
Attention
multi-head latent attention (MLA). kv_lora_rank=512. RoPE θ=50,000.
Feed-forward / MoE
It is a sparse mixture-of-experts with 384 routed experts and top-8 routing; about 32.8B parameters are active per token.
Field-level comparison
Compared against the previous model of the same lab; where none exists, against the structurally closest model. The ratio column is this model divided by the comparison model.
Full-field comparison vs predecessor Moonlight 16B
| Model | Kimi K2 Base | Moonlight 16B | Ratio |
|---|---|---|---|
| model_type | kimi_k2 | deepseek_v3 | ≠ |
| architectures | DeepseekV3ForCausalLM | DeepseekV3ForCausalLM | ＝ |
| hidden_size | 7168 | 2048 | ×3.50 |
| num_hidden_layers | 61 | 27 | ×2.26 |
| num_attention_heads | 64 | 16 | ×4.00 |
| num_key_value_heads | 64 | 16 | ×4.00 |
| intermediate_size | 18432 | 11264 | ×1.64 |
| hidden_act | silu | silu | ＝ |
| n_routed_experts | 384 | 64 | ×6.00 |
| num_experts_per_tok | 8 | 6 | ×1.33 |
| n_shared_experts | 1 | 2 | ×0.500 |
| moe_intermediate_size | 2048 | 1408 | ×1.45 |
| kv_lora_rank | 512 | 512 | ≈1 |
| qk_rope_head_dim | 64 | 64 | ≈1 |
| q_lora_rank | 1536 | — | |
| rope_theta | 50000 | 50000 | ≈1 |
| max_position_embeddings | 131072 | 8192 | ×16.00 |
| rms_norm_eps | 0.000001 | 0.00001 | ×0.100 |
| vocab_size | 163840 | 163840 | ≈1 |
| torch_dtype | bfloat16 | bfloat16 | ＝ |
| num_nextn_predict_layers | 0 | 0 | ＝ |
| attention_bias | false | false | ＝ |
| attention_dropout | 0 | 0 | ＝ |
| aux_loss_alpha | 0.001 | 0.001 | ≈1 |
| bos_token_id | 163584 | 163584 | ≈1 |
| eos_token_id | 163585 | 163585 | ≈1 |
Most similar architectures
- Kimi K2 0905 (Kimi, 2025-09) 1.000
- Kimi K2.5 (Kimi, 2026-02) 1.000
- Kimi K2.6 (Kimi, 2026-04) 1.000
- DeepSeek-V3 (DeepSeek, 2024-12) 1.000
- DeepSeek-V3 Base (DeepSeek, 2024-12) 1.000
- DeepSeek-R1 (DeepSeek, 2025-01) 1.000
Similarity values range from 0 (no shared categorical features) to 1 (identical profiles).
Raw config fields
42 fields
| architectures | DeepseekV3ForCausalLM |
|---|---|
| attention_bias | false |
| attention_dropout | 0 |
| aux_loss_alpha | 0.001 |
| bos_token_id | 163584 |
| eos_token_id | 163585 |
| first_k_dense_replace | 1 |
| hidden_act | silu |
| hidden_size | 7168 |
| initializer_range | 0.02 |
| intermediate_size | 18432 |
| kv_lora_rank | 512 |
| max_position_embeddings | 131072 |
| model_type | kimi_k2 |
| moe_intermediate_size | 2048 |
| moe_layer_freq | 1 |
| n_group | 1 |
| n_routed_experts | 384 |
| n_shared_experts | 1 |
| norm_topk_prob | true |
| num_attention_heads | 64 |
| num_experts_per_tok | 8 |
| num_hidden_layers | 61 |
| num_key_value_heads | 64 |
| num_nextn_predict_layers | 0 |
| pretraining_tp | 1 |
| q_lora_rank | 1536 |
| qk_nope_head_dim | 128 |
| qk_rope_head_dim | 64 |
| rms_norm_eps | 0.000001 |
| rope_theta | 50000 |
| routed_scaling_factor | 2.827 |
| scoring_func | sigmoid |
| seq_aux | true |
| tie_word_embeddings | false |
| topk_group | 1 |
| topk_method | noaux_tc |
| torch_dtype | bfloat16 |
| transformers_version | 4.48.3 |
| use_cache | true |
| v_head_dim | 128 |
| vocab_size | 163840 |