IFM · Released Sep 1, 2026 · observed 31 Oct 2026
Public-source support evidence
We ran the Q4_K_M GGUF on one Mac; see Hardware notes below. Community MLX weights are available; the architecture is listed in llama.cpp and vLLM.
IFM/K2-Horizon-MoVA-36B-A4B
K2-Horizon-MoVA-36B-A4B is a text generation model published by IFM on Hugging Face. We have not written a summary yet; see the model card ↗.
View source model ↗SUPPORT EVIDENCE
The model config architecture is listed in this runtime registry. This is family/architecture evidence only; exact checkpoint loading, model-specific configuration, hardware, and generation were not tested.
Checked 2026-10-31 01:03 UTC
Evidence source ↗The exact model and runtime-specific sources were checked; no positive model-level evidence was established.
Checked 2026-10-31 01:02 UTC
Version v0.40.13
Checked source ↗The exact model and runtime-specific sources were checked; no positive model-level evidence was established.
Checked 2026-10-31 01:02 UTC
Checked source ↗A current public MLX upload with an exact base_model match was observed. Upload availability only; this model was not loaded or hardware-tested.
Checked 2026-10-31 01:03 UTC
Evidence source ↗The model config architecture is listed in this runtime registry. This is family/architecture evidence only; exact checkpoint loading, model-specific configuration, hardware, and generation were not tested.
Checked 2026-10-31 01:03 UTC
Evidence source ↗MLX community uploads may target mlx-vlm; an upload observation does not establish mlx-lm compatibility.
MEMORY ESTIMATES
The summary estimate of about 29 GB is the recommended Q4_K_M quant (22.4 GB observed), from file size × 1.2 + 1.5 GB. Each card is the largest observed quant whose estimate fits that capacity.
Which file should I grab?
IQ2_S · 12.0 GB observed · about 16 GB ↗About 16 GB using file size × 1.2 + 1.5 GB. Fit is an estimate; compatibility, speed, context length and hardware have not been tested.
Which file should I grab?
IQ3_M · 17.9 GB observed · about 24 GB ↗About 24 GB using file size × 1.2 + 1.5 GB. Fit is an estimate; compatibility, speed, context length and hardware have not been tested.
Which file should I grab?
Q4_1 · 23.8 GB observed · about 31 GB ↗About 31 GB using file size × 1.2 + 1.5 GB. Fit is an estimate; compatibility, speed, context length and hardware have not been tested.
Which file should I grab?
Q8_0 · 39.8 GB observed · about 50 GB ↗About 50 GB using file size × 1.2 + 1.5 GB. Fit is an estimate; compatibility, speed, context length and hardware have not been tested.
Which file should I grab?
BF16 · 74.9 GB observed · about 92 GB ↗About 92 GB using file size × 1.2 + 1.5 GB. Fit is an estimate; compatibility, speed, context length and hardware have not been tested. Link points to shard 1; all 2 linked shards are required.
Which file should I grab?
BF16 · 74.9 GB observed · about 92 GB ↗About 92 GB using file size × 1.2 + 1.5 GB. Fit is an estimate; compatibility, speed, context length and hardware have not been tested. Link points to shard 1; all 2 linked shards are required.
Which file should I grab?
BF16 · 74.9 GB observed · about 92 GB ↗About 92 GB using file size × 1.2 + 1.5 GB. Fit is an estimate; compatibility, speed, context length and hardware have not been tested. Link points to shard 1; all 2 linked shards are required.
Which file should I grab?
BF16 · 74.9 GB observed · about 92 GB ↗About 92 GB using file size × 1.2 + 1.5 GB. Fit is an estimate; compatibility, speed, context length and hardware have not been tested. Link points to shard 1; all 2 linked shards are required.
Memory estimates are the observed file size plus fixed headroom (20% plus 1.5 GB runtime allowance); real needs also depend on context length, KV cache and runtime settings. Fit is an estimate; it does not establish runtime compatibility, speed, context length, or hardware performance. Text-weights-only estimates do not include a projector or total vision memory where unknown.
Terms: a quant is a lower-precision version of the weights; GGUF is a file format used by llama.cpp and compatible apps; MLX targets Apple silicon; vLLM is a model-serving engine for NVIDIA GPUs; the KV cache is stored attention state that grows with context length. Terms explained ↗
OBSERVED FILES
Sizes and file groups come from the listed uploader repositories. A size is marked as an estimated fit only for a complete observed GGUF quant group.
| Uploader | Format & quant | Observed size | Estimated fit | Files |
|---|---|---|---|---|
| IFMIFM/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · BF16 | 74.9 GB | 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file
|
| IFMIFM/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q4_K_M | 22.4 GB | 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| IFMIFM/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q5_0 | 25.9 GB | 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| IFMIFM/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q5_K_M | 26.4 GB | 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| IFMIFM/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q6_K | 30.8 GB | 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| IFMIFM/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q8_0 | 39.8 GB | 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · IQ2_M | 13.1 GB | 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · IQ2_S | 12.0 GB | 16 GB, 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · IQ2_XS | 11.7 GB | 16 GB, 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · IQ2_XXS | 10.6 GB | 16 GB, 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · IQ3_M | 17.9 GB | 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · IQ3_XS | 17.3 GB | 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · IQ3_XXS | 15.8 GB | 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · IQ4_NL | 21.5 GB | 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · IQ4_XS | 20.4 GB | 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q2_K | 13.6 GB | 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q3_K_L | 17.8 GB | 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q3_K_M | 17.3 GB | 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q3_K_S | 16.8 GB | 24 GB, 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q4_0 | 21.6 GB | 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q4_1 | 23.8 GB | 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q4_K_L | 23.2 GB | 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q4_K_M | 22.7 GB | 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q4_K_S | 22.0 GB | 32 GB, 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q5_K_M | 26.7 GB | 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q5_K_S | 26.0 GB | 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q6_K | 32.1 GB | 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q6_K_L | 32.4 GB | 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · Q8_0 | 39.8 GB | 64 GB, 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | 1 file |
| bartowskibartowski/K2-Horizon-MoVA-36B-A4B-GGUF ↗Observed 31 Oct 2026 | GGUF · BF16 | 74.9 GB | 128 GB, 192 GB, 256 GB, 512 GBEstimate only; requires the complete GGUF quant group. | Complete observed GGUF quant group; all 2 shard(s) are required. |
| mlx-communitymlx-community/K2-Horizon-MoVA-36B-A4B-4bit ↗Observed 31 Oct 2026 | MLX · Quant unknown | Size unknown | UnknownUnknown: completeness or a complete GGUF quant group was not verified. | Observed weight file metadata; completeness unavailable without index/config verification. |
| mlx-communitymlx-community/K2-Horizon-MoVA-36B-A4B-8bit ↗Observed 31 Oct 2026 | MLX · Quant unknown | Size unknown | UnknownUnknown: completeness or a complete GGUF quant group was not verified. | 8 files
|
| mlx-communitymlx-community/K2-Horizon-MoVA-36B-A4B-oQ4e ↗Observed 31 Oct 2026 | MLX · Quant unknown | Size unknown | UnknownUnknown: completeness or a complete GGUF quant group was not verified. | Observed weight file metadata; completeness unavailable without index/config verification. |
CURATED TRACKING
No curated issues verified for this snapshot. An empty list does not mean there are no bugs.
HARDWARE NOTES
We ran the Q4_K_M GGUF (22.37 GB, IFM/K2-Horizon-MoVA-36B-A4B-GGUF @00cce31) on an Apple M4 Max Mac, 128 GiB unified memory, macOS 27.0.1, Metal with llama.cpp b11912 (upstream, 961c376) on 28 Oct 2026.
Decode: 71.9 tok/s on a short prompt, 68.6 tok/s on a long prompt (medians of 3 runs).
Peak memory: 24.0 GiB (llama-server footprint).
llama-bench: pp512 1,043.8 tok/s, tg128 73.5 tok/s.
Answer checks: 2 of 4 prompts passed (quick pass/fail sanity checks, not a quality benchmark).
We ran IFM's official Q4_K_M GGUF of this model.
Default chat template settings (no thinking toggle passed). The model wrote reasoning text on every prompt; decode speed counts every generated token.
Both failed checks hit the 512-token output cap: one while the model was still reasoning, one partway through the answer.
One machine, one run set: temperature 0, 8K context, 1 warm-up plus 3 runs per prompt. Short prompt is 20 tokens; long prompt is 1,398 tokens (each model counts with its own tokenizer). Weights were loaded from an external USB drive with mmap off; load time is not published for this run. Your speed will differ with hardware, context length and settings.
COMPARE
No comparison page is generated for this model yet. Browse all comparisons →
EMBED
Works in GitHub READMEs and Hugging Face model cards. The badge updates when the site rebuilds.