Kimi-K2.5

2026-01-27

Date the weights (or, if earlier, the technical report) were first made public, YYYY-MM-DD, as dated by the publisher. A Hugging Face repository creation date is used only when no dated publisher page exists, with a note.

other (modified-mit)

License of the released weights as declared by the publisher (model card metadata or license file), e.g. apache-2.0, mit, or the name of a custom license.

text + image + video

Input modalities of the released checkpoint as the publisher states them, from: text, image, video, audio, joined with ' + ' in that order ("text", "text + image + video").

1T parameters

Total parameter count as reported by the publisher. See params.note for whether embeddings, MTP heads or vision towers are included.

32B parameters

Parameters used per token in the forward pass, as reported by the publisher.

Not disclosed

Active parameters excluding the input embedding and the output head (unembedding). Stated by the publisher, or computed from params.active by subtracting vocab × d_model for each of the embedding and output head that params.active is stated to include (basis inferred, note with the arithmetic). Not disclosed when the scope of params.active is not stated.

1.026T parameters

Calculated: embedding + output head + all experts + other weights, moonshotai/Kimi-K2.5@4d01dfe safetensors headers; reproduces stated total (backbone) and active (with output head)

Parameters of the text backbone counted from the released checkpoint's safetensors headers (embedding, output head, all experts and other weights; excluding MTP modules, vision/audio towers and n-gram lookup tables). Shown only when the counts reproduce the publisher's stated total and active parameters within 5% under some scope.

2.349B parameters

Calculated: input embedding + output head, moonshotai/Kimi-K2.5@4d01dfe safetensors headers

Input embedding plus output head (once if tied), counted from the checkpoint.

1.024T parameters

Calculated: dense + routed backbone weights; excludes input embedding, output head, lookup tables, MTP and other modalities, moonshotai/Kimi-K2.5@4d01dfe safetensors headers; reproduces stated total (backbone) and active (with output head)

Language backbone weights excluding input embeddings, output head, lookup/Engram tables, MTP modules and vision/audio towers. Equals params.counted_total minus params.counted_embedding. Same validation as params.counted_total.

17.06B parameters

Calculated: max(sum of dense and routed weights sharing a full layers.<index> tensor path), all experts included, moonshotai/Kimi-K2.5@4d01dfe safetensors headers

Largest group of language-backbone weights sharing a full layers.<index> tensor path, including every routed expert. Excludes lookup/Engram tables, embeddings, output head, MTP and other modalities. Hybrid checkpoints may store mixers and FFNs as separate layers. All-gather volume also depends on wrapping, sharding and dtype.

30.51B parameters

Calculated: non-expert weights + routed experts × 8/384, excluding embedding and output head, moonshotai/Kimi-K2.5@4d01dfe safetensors headers; reproduces stated total (backbone) and active (with output head)

Active parameters per token excluding the input embedding and output head, counted from the checkpoint as all non-expert weights plus routed-expert weights × top_k / experts. Same validation as params.counted_total.

Rounded card figures; whether they include the 400M MoonViT encoder is not stated.

What the publisher's counts include or exclude (MTP modules, embeddings, vision encoder, n-gram tables).

61

Number of transformer blocks in the main backbone (attention or token mixer plus FFN), excluding MTP modules.

7,168

Residual stream width.

163,840

Embedding table rows (vocab_size in config, may include padding).

No

Input embedding and output head share weights.

117.508

Calculated: d_model / layers

Width-to-depth ratio, d_model divided by the number of blocks.

MLA

Token-mixing layer types in the backbone, e.g. GQA, MLA, SWA, Gated DeltaNet, Mamba-2, DSA (DeepSeek sparse attention).

61 MLA

Token-mixer order, summarized as a pattern or layer positions. The fact's optional sequence lists each mixer in order; a ratio alone does not establish the order. Auxiliary prediction modules and standalone FFNs are excluded.

61

Layers whose KV cache grows with full context (dense or sparse global attention, MLA). Excludes sliding-window and linear/recurrent layers.

64

Query heads in the global-attention layers.

12,288

Calculated: heads × head_dim

attn.heads × attn.head_dim, the query projection width of global layers.

192

Query/key dimension per head in global-attention layers (for MLA, nope + rope dims).

128

Value dimension per head in global layers (MLA v_head_dim; stated for MLA even when equal to the QK head dim).

512

Width of the compressed KV latent that MLA caches per token per layer.

1,536

Width of the low-rank query projection in MLA; 0 when queries use a full-rank projection.

64

Width of the shared positional key that MLA caches next to the latent.

partial RoPE (MLA decoupled, 64 of 192 QK dims)

Positional scheme in global layers: RoPE, partial RoPE, NoPE, or mixed per layer type.

50,000

RoPE base frequency of the global layers at release.

No

Queries and keys are normalized per head before the dot product.

No

Query/key/value projections have bias terms.

SwiGLU

FFN activation (SwiGLU, GeGLU, ...).

18,432

Intermediate width of the dense FFN layers that are actually executed.

2.57143

Calculated: ffn.dim / d_model

ffn.dim divided by shape.d_model.

384

Routed experts per MoE layer.

8

Routed experts selected per token.

1

Always-on experts per MoE layer.

2,048

Intermediate width of one routed expert.

2,048

Inferred from the source; see the explanation below.

Total intermediate width of the shared expert(s).

1

Number of initial blocks that use a dense FFN instead of MoE.

sigmoid

Gating function over expert logits (softmax, sigmoid).

aux-loss-free selection bias (e_score_correction_bias added to sigmoid scores for top-k only)

Load-balancing method, with coefficients where disclosed.

2.827

Multiplier applied to the combined routed-expert output (routed_scaling_factor).

Yes

Selected expert weights are renormalized to sum to 1.

none

Expert groups and groups selected per token, or node-limited routing, when used.

0.285714

Calculated: moe.expert_dim / d_model

moe.expert_dim divided by shape.d_model.

48

Calculated: experts / top_k

Routed experts divided by experts selected per token.

0.032

Calculated: params.active / params.total

params.active divided by params.total.

Not disclosed

Additional future tokens predicted during pretraining (0 when the model was trained without MTP).

No

The released checkpoint contains the MTP module weights.

RMSNorm

Normalization layer type in the blocks.

pre

Pre-norm, post-norm, or sandwich (pre and post).

1e-05

Epsilon of the block normalization layers.

4.096K tokens

Sequence length of the main pretraining stage.

262.1K tokens

Longest sequence length trained on after context extension.

262.1K tokens

Longest context the publisher says the model supports, possibly beyond trained length via RoPE scaling.

YaRN, 32K → 256K during long-context mid-training (500B → 200B tokens)

How context was extended (stages, RoPE base change, YaRN factor).

15T tokens

Tokens seen in pretraining including annealing and context extension, as reported.

Not disclosed

Optimizer used for the bulk of parameters.

Not disclosed

Peak learning rate of the main pretraining stage.

Not disclosed

Learning rate at the end of the decay.

Not disclosed

Learning-rate schedule shape with phase boundaries.

Not disclosed

Warmup length in steps or tokens.

Not disclosed

Global batch size, including any batch-size ramp: sequences × sequence length (token total), or the token total alone when only that is stated.

Not disclosed

Decoupled weight decay coefficient.

Not disclosed

(beta1, beta2) of the Adam-family optimizer or its AdamW fallback.

Not disclosed

Global gradient-norm clipping threshold.

Not disclosed

Matmul precision of pretraining (BF16, FP8 recipe, NVFP4 recipe) with the scaling granularity where disclosed. Never inferred from the released checkpoint dtype.

Language model continued from a near-end Kimi K2 checkpoint; MoonViT-3D from SigLIP-SO-400M

Weight initialization (std or scheme) as stated by the publisher; config initializer_range alone is not evidence.

Joint vision-text next-token prediction with a constant vision ratio from the start of continued pretraining

Objectives beyond next-token prediction and MTP (fill-in-the-middle rate, distillation, multimodal tokens from the start).

Kimi K2 distribution extended with new unique tokens, higher coding weight and per-source epoch caps, plus interleaved image, video and OS-screenshot data

Pretraining data composition as disclosed (domain shares, synthetic fraction, repetition).

NVIDIA H800, 8×400 Gbps RoCE between nodes

Accelerators used for pretraining, count and duration where stated.

H800

Accelerator count and type for the main pretraining run, as the count with thousands separators and the chip's short name ("2,048 H800", "2,048 GH200"). A stated range or bound stays as stated ("2,048–4,096 GH200").

PP16 · VPP · EP16 · ZeRO-1 DP

Inferred from the source; see the explanation below.

Parallelism layout of the main pretraining run as stated, as degree-annotated terms in the order TP, PP, VPP, EP, CP, DP ("PP16 · EP64 · ZeRO-1 DP"); a term without a stated degree has none ("FSDP").

• vit

  Tokens

  1000000000000

  Seq len

  4096

  Source: paper · p.7 Table 3; p.7–8 §4.3

  consuming about 1T tokens with very few training FLOPs.

  MoonViT-3D only, caption loss aligned to Moonlight-16B-A3B; then a short projector-only stage.

• joint

  Tokens

  15000000000000

  Seq len

  4096

  Source: paper · p.7 Table 3; p.7–8 §4.3

  over additional 15T vision-text tokens at 4K sequence length.

  ViT and LLM trained.

• long-context

  Seq len

  262144

  Source: paper · p.7 Table 3; p.7–8 §4.3

  500B→200B

  Two phases, 500B tokens at 32K then 200B at 256K; high-quality text, multimodal, long text, long video, reasoning and long-CoT data.

Ordered pretraining stages. Each item: name, tokens, seq_len, lr, batch_tokens, batch_sequences, steps, note, quote, and source when it differs from the fact's source.

15

Calculated: train.tokens / params.total

train.tokens divided by params.total.

491.599

Calculated: train.tokens / params.counted_active_non_embedding

train.tokens divided by params.counted_active_non_embedding, or by the stated params.active_non_embedding when there is no validated checkpoint count.

70,272 bytes

Calculated: kv.bytes_at_context / 131072 tokens; uniform BF16, payload only; quantization overhead excluded. Global KV: 61 × (576 × 16/8) B/entry

Retained KV cache divided by context length, including bounded SWA caches and indexer keys. Exports use uniform BF16 at 131,072 tokens. The website can select uniform payload precision (no quantization overhead) or sourced component formats (including scales and documented padding). This is cache payload, not peak prefill memory or persistent prefix storage.

70,272 bytes

Calculated: 70272 B / 1 tokens (Global KV); uniform BF16, payload only; quantization overhead excluded. Global KV: 61 × (576 × 16/8) B/entry

Long-context global KV bytes per token, including indexer keys; excludes bounded SWA caches. Compression rates and cross-layer cache reuse are included. Exports use uniform BF16; the website's storage selector also applies here. This is the field to compare with global-KV figures in reports, using the matching storage recipe.

2,196 bytes

Calculated: kv.bytes_per_token (131072 tokens) / (params.active / 1e9); uniform BF16, payload only; quantization overhead excluded. Global KV: 61 × (576 × 16/8) B/entry

kv.bytes_per_token divided by active parameters in billions (131,072 tokens in exports).

287,834,112 bytes

Calculated: kv.bytes_at_context (131072 tokens) / (params.active / 1e9); uniform BF16, payload only; quantization overhead excluded. Global KV: 61 × (576 × 16/8) B/entry

kv.bytes_at_context divided by active parameters in billions (131,072 tokens in exports).

9,210,691,584 bytes

Calculated: 70272 B × 131072 (Global KV); uniform BF16, payload only; quantization overhead excluded. Global KV: 61 × (576 × 16/8) B/entry

Retained KV cache for one sequence at the selected context, including SWA windows and indexer keys. Exports use uniform BF16 at 131,072 tokens. Excludes recurrent state, compressor tail state, draft/MTP caches, temporary prefill tensors, allocator metadata and persistent prefix snapshots. Does not estimate peak prefill or total serving memory.

23

Artificial Analysis Intelligence Index v4.3.2 of the post-trained model, from a dated Wayback snapshot of its Artificial Analysis page. The note names the variant (reasoning setting).