llmrun
Thinking MachinesInkling 2Inkling2LoopForConditionalGeneration

Inkling 2 Mini — Hardware Requirements & GPU Compatibility

VisionChatFunctionsReasoning

Inkling 2 Mini is an Apache 2.0 vision-language model from Thinking Machines Lab, released in November 2026. It has 26.4 billion physical parameters and activates about 3.4 billion per token. Its middle twelve layers run three times, producing about 5.4 billion effective weight activations per token without storing three copies of those layers. It accepts text and images, produces text, and supports up to one million tokens of context. The developer distributes a mixed NVFP4 and MXFP8 checkpoint of about 20 GB and recommends Apple silicon with at least 32 GB of memory, an NVIDIA DGX Spark, or other supported hardware with sufficient memory. GGUF conversions are available for compatible llama.cpp and Ollama runtimes. Repeated layers increase computation per token, so measured throughput and memory use in the chosen runtime matter more than the active-parameter count alone.

26.4B params · MoE3.4B active1049K contextApache 2.0136 downloads1.0K likes
S

Runs great on a 24 GB GPU

Runs from12 GB IQ2_XXS · 16KBest on24 GB Q4_K_M · ~211 tok/s · 16KMacs from16 GB IQ2_XXS · ~82 tok/s

Apache 2.0

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run Inkling 2 Mini

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Recommended

Ollama
one command · needs Ollama installed
$ollama run hf.co/lmstudio-community/Inkling-2-Mini-GGUF:Q4_K_M

Q4_K_M fits fully on the RTX 4090 · 3.0 GB free · 16K context

Download: Inkling-2-Mini-Q4_K_M.gguf17.48 GBlmstudio-community

install, OS, other quants

Ways to runOther ways

each opens step-by-step

Run in cloud

Runs on RTX PRO 4500 (32 GB) (11 GB headroom) · Q4_K_M

Generation speed
~193 tok/s
generation speed
Cost per 1M output tokens
$0.49
per 1M output tokens

How Much VRAM Does Inkling 2 Mini Need?

Select a quantization level; the listed hardware adheres to it.

QuantizationVRAM+32KOn RTX 4090 (reference)
12.3 GB20 GB2-bit · K-quant — 2-bit quantization with K-quant improvementsFits: 32K context · ~267 tok/s
12.7 GB20.4 GB3-bit small — 3-bit small quantizationFits: 32K context · ~262 tok/s
14.2 GB21.8 GB3-bit medium — 3-bit medium quantizationFits: 32K context · ~244 tok/s
14.5 GB22.2 GB4-bit · legacy — 4-bit legacy quantizationFits: 32K context · ~240 tok/s
17.4 GB25.1 GB4-bit medium — 4-bit medium quantization — most popular sweet spotFits: 16K context · ~211 tok/s
20.7 GB28.4 GB5-bit medium — 5-bit medium quantization — good quality/size tradeoffFits: 8K context · ~186 tok/s
24.0 GB31.6 GB6-bit, very good — 6-bit quantization, very good qualityFits: 2K context · ~166 tok/s
29.0 GB36.7 GB8-bit, near-lossless — 8-bit quantization, near-losslessDoes not fit: needs 32.6 GB

Selected Q4_K_M

4-bit medium quantization

21.0GBat 16K context

Weights
16.6 GB
KV cache
4.1 GB
Overhead
0.3 GB

Fits RTX 4090 (reference) · 3.0 GB free · ~211 tok/s

≥50 tok/s needs ≥185 GB/s · RTX 5090 ~308, RTX 3090 Ti ~211

Which Hardware Can Run Inkling 2 Mini?

Q4_K_M · 17.4 GB

Inkling 2 Mini (Q4_K_M) requires 17.4 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 23+ GB is recommended. 8 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 137 devices with unified memory can run Inkling 2 Mini, including Mac Studio M5 Ultra (256 GB), Mac Studio M5 Ultra (512 GB), NVIDIA RTX Spark N1X Laptop (32 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M
8 GB · RTX 4060Doesn't fit — smallest quant needs 11.6 GB8 GB tier →
12 GB · RTX 3060 12GBIQ2_XXS · 11.6 GB · ~175 tok/s · 16K12 GB tier →
16 GB · RTX 5060 Ti 16GBIQ3_S · 15.9 GB · ~148 tok/s · 16K16 GB tier →
24 GB · RTX 4090referenceQ4_K_M · 21 GB · ~211 tok/s · 16K24 GB tier →
Mac 32 GB · Mac Mini M4Q4_K_M · 21 GB · ~32 tok/s · 32KMacs →
Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 21 GB · ~94 tok/s · 128KMacs →

Alternatives to Inkling 2 Mini

Similar models

6
ModelScoreVRAMtok/sWhy it is similar
Qwen3.8 27BAlibaba121.917.9 GB~38same size
Qwen3.6 27BAlibaba111.217.9 GB~38same size
Qwen3.5 27BAlibaba103.817.9 GB~38same size
Gemma 3 27B ITGoogle79.421.7 GB~36same size
Gemma 4 26B A4BGoogle—16.5 GB~171same size
Ariarhymes-ai—20.2 GB~168same size

Specifications

Publisher
Thinking Machines
Family
Inkling 2
Parameters
26.4B
Active parameters
3.4B per token
Architecture
Inkling2LoopForConditionalGeneration
Context Length
1,048,576 tokens
Release Date
2026-11-20
License
Apache 2.0

Frequently Asked Questions

How much VRAM does Inkling 2 Mini need?

Inkling 2 Mini requires 17.4 GB of VRAM at Q4_K_M, or 58.1 GB at BF16.

VRAM = Weights + KV Cache + Overhead

Weights = 26.4B × 4.8 bits ÷ 8 = 15.8 GB

KV Cache + Overhead ≈ 1.6 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

VRAM usage by quantization

17.4 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 4090 run Inkling 2 Mini?

Yes, at Q6_K (24.0 GB) or lower. Higher quantizations like Q8_0 (29.0 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.

What's the best quantization for Inkling 2 Mini?

For Inkling 2 Mini, Q4_K_M (17.4 GB) offers the best balance of quality and VRAM usage. Q4_K_L (17.8 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 8.0 GB.

VRAM requirement by quantization

IQ2_XXS
8.0 GB
Q2_K
12.3 GB
Q3_K_L
14.9 GB
Q4_K_M ★
17.4 GB
Q4_K_L
17.8 GB
BF16
58.1 GB

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run Inkling 2 Mini on a Mac?

Yes — Mac Mini M2 Pro (16 GB) and 145 other Macs can run Inkling 2 Mini, the smallest at IQ2_XXS. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 17.4 GB of usable unified memory (RAM minus macOS overhead), from Mac Mini M5 Pro (24 GB).

Can I run Inkling 2 Mini locally?

Yes — Inkling 2 Mini can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is Inkling 2 Mini?

At Q4_K_M, Inkling 2 Mini can reach ~560 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~211 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)

Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.

Example: NVIDIA B200 → 2.2 GB active ÷ (8000 × 0.65) = 0.43 ms, plus 24 layers × 0.055 ms = 1.32 ms, so 1000 ÷ 1.75 ms = ~571 tok/s

Estimated speed at Q4_K_M (17.4 GB)

~571 tok/s
~211 tok/s
~571 tok/s
~560 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of Inkling 2 Mini?

At Q4_K_M, the download is about 15.84 GB. The full-precision BF16 version is 52.80 GB. The smallest option (IQ2_XXS) is 7.26 GB.

Which GPUs can run Inkling 2 Mini?

8 consumer GPUs can run Inkling 2 Mini at Q4_K_M. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XT, AMD Radeon RX 7900 XTX. 1 GPU has plenty of headroom for comfortable inference.

Which devices can run Inkling 2 Mini?

137 devices with unified memory can run Inkling 2 Mini at Q4_K_M (17.4 GB), including AMD Ryzen 7 8845HS (Hawk Point) Laptop, AMD Ryzen AI 7 350 (Krackan Point) Laptop, AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10. Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.