Step 3.7 Flash 196B-A11B VRAM requirements

Step 3.7 Flash 196B-A11B needs about 126 GB of VRAM at Q4_K_M with 8K context, so it needs a 141 GB card such as the H200; a 24 GB card is too small. Step 3.7 Flash 196B-A11B has 201.4B parameters (the 196B in the name is the language model; the checkpoint's parameter count includes weights outside it, such as MTP layers), of which about 11B are used per token; all 288 experts still have to be in memory. With the same 8K context and one request it needs 207 GB at FP8 and 414 GB at FP16/BF16. The published weights take 375 GB (BF16). It does not fit on a 24 GB card at Q4_K_M; the smallest setup here that holds it with an 8K context is the H200 (141 GB).

126 GB at Q4_K_M, 8K context, one request

FP8
207 GB
FP16 / BF16
414 GB
Published weights
375 GB
Smallest setup, Q4_K_M
H200 (141 GB)
Open Step 3.7 Flash 196B-A11B in the calculator

Worked example: Step 3.7 Flash 196B-A11B with 32K tokens

Inputs: Step 3.7 Flash 196B-A11B · Q4_K_M weights · 32,768 tokens of context · one request · FP16 KV cache

Weights (201.4B at Q4_K_M)
113 GB
KV cache (48 KB per token)
1.63 GB
Buffers and runtime (0.5 GB + 10%)
12.0 GB
Total
127 GB

The smallest setup here that holds it is the H200 (141 GB), with about 13.9 GB to spare. Change the inputs in the calculator

What makes Step 3.7 Flash 196B-A11B's memory use different

Its 33 sliding-window layers keep only the last 512 tokens (llama.cpp gives them 1,024 cells: the window plus a 512-token batch, rounded up to 256): at its 256K limit they hold 132 MB of cache instead of the 33.0 GB they would need with full attention.

Per token of context it adds 48 KB of FP16 cache; Mistral Medium 3.5 128B (127.7B), the nearest-sized model here with plain full attention, adds 352 KB, so Step 3.7 Flash 196B-A11B needs 14% as much.

As a mixture-of-experts model it reads about 6.20 GB of its 113 GB Q4_K_M weights per generated token (5%), so it writes like a much smaller model while needing memory for all of them.

How much VRAM does Step 3.7 Flash 196B-A11B need?

Step 3.7 Flash 196B-A11B needs 126 GB at Q4_K_M, 220 GB at Q8_0 and 414 GB at FP16 with 8,192 tokens of context and one request. Each total below is the weights plus the FP16 KV cache and the runtime overhead (0.5 GB plus 10%), and each row opens the calculator with that setting. What the GGUF names mean.

Fine-tuning Step 3.7 Flash 196B-A11B? Step 3.7 Flash 196B-A11B VRAM for LoRA, QLoRA and full training.

Step 3.7 Flash 196B-A11B GGUF files on Hugging Face

The 8 GGUF files llama.cpp picks for Step 3.7 Flash 196B-A11B with -hf <repo>:<QUANT>, with the bytes the Hugging Face file list reported on 2026-09-29. Each total is that file plus the FP16 KV cache and 0.5 GB + 10%, one request. "vs estimate" compares the file with the bits-per-weight size the table above uses.

QuantFileSizevs estimate 8K32K128K
Q2_K bartowski/Step-3.7-Flash-GGUF2 parts 66.3 GB71,226,835,104 bytes −16% 74.0 GB75.3 GB80.2 GB
Q3_K_M unsloth/Step-3.7-Flash-GGUF3 parts 83.1 GB89,262,719,488 bytes −9% 92.5 GB93.7 GB98.7 GB
IQ4_XS unsloth/Step-3.7-Flash-GGUF3 parts 88.8 GB95,336,010,208 bytes −13% 98.7 GB100.0 GB105 GB
Q4_K_M unsloth/Step-3.7-Flash-GGUF4 parts 114 GB122,090,426,976 bytes ±0% 126 GB127 GB132 GB
Q5_K_M unsloth/Step-3.7-Flash-GGUF4 parts 136 GB146,495,447,648 bytes +3% 151 GB152 GB157 GB
Q6_K unsloth/Step-3.7-Flash-GGUF5 parts 152 GB163,340,821,184 bytes −1% 168 GB170 GB175 GB
Q8_0 unsloth/Step-3.7-Flash-GGUF6 parts 195 GB209,417,872,192 bytes −2% 216 GB217 GB222 GB
BF16 unsloth/Step-3.7-Flash-GGUF8 parts 367 GB394,017,424,160 bytes −2% 405 GB406 GB411 GB

GB here is GiB (1024³ bytes), as everywhere on this site; Hugging Face shows the Q4_K_M file as 122.1 GB, in 10⁹ bytes.

Which Step 3.7 Flash 196B-A11B GGUF fits a 16, 24, 32, 48 or 80 GB GPU or a Mac

From the file sizes above: the largest file that fits each machine with an 8K context and 0.5 GB left free, the longest context it then has room for, and the longest context for Q4_K_M (114 GB). One request, FP16 KV cache, -np 1.

HardwareLargest GGUF, 8KIts longest contextQ4_K_M, longest context
RTX 5060 Ti 16GB None fits — No
RTX 4090 (24 GB) None fits — No
RTX 5090 (32 GB) None fits — No
L40S (48 GB) None fits — No
H100 SXM (80 GB) Q2_K, 66.3 GB 114K No
M4 Pro Mac (64 GB, 48 GB usable) None fits — No
M4 Max Mac (128 GB, 96 GB usable) Q3_K_M, 83.1 GB 65K No

Run Step 3.7 Flash 196B-A11B with llama.cpp

llama-server -hf unsloth/Step-3.7-Flash-GGUF:Q4_K_M -c 262144 -np 1

UD-Q4_K_M/Step-3.7-Flash-UD-Q4_K_M-00001-of-00004.gguf, 122.1 GB, from unsloth/Step-3.7-Flash-GGUF (checked 2026-09-29). -c 262144 is the longest Q4_K_M context on the H200 (141 GB). -np 1: one slot, so one 512-token window.

Step 3.7 Flash 196B-A11B at 8K, 32K, 128K and 256K (full) tokens of context

12 of its 45 layers use full attention and 33 keep a sliding window of 512 tokens. Each extra token of context adds 48 KB of FP16 cache per request once the sliding windows are full. The last column is the smallest setup here, one card or a group, that holds it at Q4_K_M. How the KV cache works.

ContextKV cache, FP16Total at Q4_K_MTotal at Q8_0Smallest setup, Q4_K_M
8K tokens 516 MB 126 GB 220 GB H200 (141 GB)
32K tokens 1.63 GB 127 GB 221 GB H200 (141 GB)
128K tokens 6.13 GB 132 GB 226 GB H200 (141 GB)
256K tokens 12.1 GB 139 GB 233 GB H200 (141 GB)

How fast Step 3.7 Flash 196B-A11B writes

Tokens per second for one request with 8,192 tokens of context, estimated from memory bandwidth and the parameters read per token. A dash means it does not fit on one card. Try other settings in the speed calculator.

HardwareBandwidthQ4_K_MFP8
RTX 3060 12GB 360 GB/s ——
RTX 4090 1,008 GB/s ——
RTX 5090 1,792 GB/s ——
M4 Max Mac (128 GB) 546 GB/s ——
M3 Ultra Mac Studio (512 GB) 819 GB/s 32–5521–35
H100 SXM 3,350 GB/s ——
H200 4,800 GB/s 154–286—

Longest context on one GPU

How many tokens of context Step 3.7 Flash 196B-A11B fits on a single card with one request, an FP16 KV cache and 0.5 GB left free. "Full" means the model's whole context window fits.

GPUMemoryQ4_K_MQ8_0FP8
RTX 3060 12 GB NoNoNo
RTX 4060 Ti 16GB 16 GB NoNoNo
RTX 3090 / 4090 24 GB NoNoNo
RTX 5090 32 GB NoNoNo
A100 40GB 40 GB NoNoNo
Mac, 64 GB unified memory 48 GB NoNoNo
L40S / RTX 6000 Ada 48 GB NoNoNo
A100 / H100 80GB 80 GB NoNoNo
Mac, 128 GB unified memory 96 GB NoNoNo
H200 141 GB 256K (full)NoNo
B200 180 GB 256K (full)NoNo

Model details

Parameters
201.4B (201,365,316,160); the 196B in the name is the language model; the checkpoint's parameter count includes weights outside it, such as MTP layers
Experts
288 routed experts, all loaded
Active per token
11B
Layers
12 of its 45 layers use full attention and 33 keep a sliding window of 512 tokens
Attention cache
8 KV heads × 128
Context length
262,144 tokens
Published weights
375 GB (BF16)
On Hugging Face
stepfun-ai/Step-3.7-Flash
Reported benchmarks
None published in the model card for GPQA, MMLU-Pro or LiveCodeBench.

Why this estimate looks this way

At Q4_K_M and an 8K-token context, Step 3.7 Flash 196B-A11B uses 113 GB for weights, 516 MB for its FP16 KV cache and 11.9 GB for estimated runtime overhead, totaling 126 GB. The overhead is a 0.5 GB base plus 10% of weights and cache.

This is a mixture-of-experts model: all 288 experts contribute to the 201.4B parameters held in memory, even though only about 11B parameters run per token. Using only active parameters would understate VRAM.

Its attention layout matters for long context: 12 of its 45 layers use full attention and 33 keep a sliding window of 512 tokens. The FP16 cache grows by about 48 KB per additional token per request once the sliding windows are full.

The architecture and published checkpoint size come from the model's config.json and weight files. These are estimates rather than measured peak memory; inference engines can reserve extra buffers or preallocate the full configured cache.

Step 3.7 Flash 196B-A11B next to similar-sized models

The three models here closest to it in parameter count, at Q4_K_M with 8,192 tokens of context.

ModelParametersTotalKV per tokenSmallest setup
Step 3.7 Flash 196B-A11B 201.4B, 11B active 126 GB 48 KB H200 (141 GB)
Qwen3.8 Flash Next 180.0B, 6B active 112 GB 24 KB DGX Spark (128 GB, 120 GB usable)
MiniMax M2.7 228.7B, 10B active 144 GB 248 KB B200 (180 GB)
DeepSeek V4 Flash 290.9B, 13B active 182 GB 86 KB M3 Ultra Mac Studio (512 GB, 384 GB usable)

Step 3.7 Flash 196B-A11B VRAM questions

How much VRAM do I need to run Step 3.7 Flash 196B-A11B locally?

About 126 GB at Q4_K_M with an 8K context and one request, by the bits-per-weight estimate, and 414 GB at BF16. The Q4_K_M GGUF UD-Q4_K_M/Step-3.7-Flash-UD-Q4_K_M-00001-of-00004.gguf in unsloth/Step-3.7-Flash-GGUF is 122,090,426,976 bytes (114 GB); with an 8K context it needs 126 GB. The smallest setup listed here that holds Q4_K_M is the H200 (141 GB).

Can I run Step 3.7 Flash 196B-A11B on a 24 GB GPU?

No. The RTX 4090 (24 GB) alone cannot hold it: the smallest listed GGUF, Q2_K at 66.3 GB, needs 74.0 GB with an 8K context.

Which quantization of Step 3.7 Flash 196B-A11B fits in 16 GB of VRAM?

The RTX 5060 Ti 16GB alone cannot hold it: the smallest listed GGUF, Q2_K at 66.3 GB, needs 74.0 GB with an 8K context.

Can I run Step 3.7 Flash 196B-A11B on a Mac?

The M4 Pro Mac (64 GB, 48 GB usable) alone cannot hold it: the smallest listed GGUF, Q2_K at 66.3 GB, needs 74.0 GB with an 8K context. On the M4 Max Mac (128 GB, 96 GB usable), the largest listed GGUF that fits with an 8K context and 0.5 GB free is Q3_K_M (83.1 GB), with up to 65K tokens of context. By default macOS lets the GPU use about 75% of unified memory.

How much more VRAM does Step 3.7 Flash 196B-A11B need for 32K or 128K tokens of context?

Its FP16 KV cache is 516 MB at 8K, 1.63 GB at 32K, 6.13 GB at 128K for one request, so 128K adds 5.63 GB over 8K, plus 10% overhead. A q8_0 cache (-ctk q8_0 -ctv q8_0) takes 3.26 GB at 128K. Its 33 sliding-window layers stop growing at 512 tokens, so only 12 layers keep adding cache.

Why does Step 3.7 Flash 196B-A11B use more VRAM than its GGUF file size?

The file holds only the weights. At Q4_K_M with an 8K context: 114 GB of weights (the file's 122,090,426,976 bytes) + 516 MB of FP16 KV cache + 11.9 GB of compute buffers and runtime (0.5 GB + 10%) = 126 GB.

Does Step 3.7 Flash 196B-A11B need less VRAM because only 11B parameters are active?

No. All 288 experts stay loaded, so the Q4_K_M file takes 114 GB. The active parameters decide speed: each generated token reads about 6.20 GB of them.

Badge for your model card

Paste it into a Hugging Face or GitHub README; it shows the Q4_K_M total above and links to this page. Q8_0 and FP16 badges.

Step 3.7 Flash 196B-A11B VRAM: 126 GB at Q4_K_M, 8K context

Numbers read from the model files on Hugging Face on . Compare it with other models in the reproducible model and GPU report.