What GPU do you need to run Thinking Machines Inkling-Small?
A new Framework Desktop ($6,799). No priced GPU card holds Thinking Machines Inkling-Small (about 160.8 GB at Q4); this system holds it with room for its full 1M context, at about 27 tok/s estimate. 1 of the 157 GPUs and systems we track fit it.
Older model Newer alternative from the same developer: Thinking Machines Inkling 2 Small (354.15 GB at UD-Q4_K_XL).
265.96B parameters, 12B active (mixture-of-experts) (Hugging Face, accessed 2026-10-18). Q4_K_M GGUF: 159.34 GB (source). Architecture config (Hugging Face config.json; local_layer_ids: 35 sliding-window (512) + 7 global layers; Inkling relative-position extras (d_rel) ignored): 7 global attention layers × 8 KV heads × 128 head dim, plus 35 sliding-window layers (512-token window), so about 28 KB of FP16 KV cache per token of context (how we estimate context).
Card: 276B total, 12B active, local + global hybrid attention. NOT supported by stock llama.cpp: the Inkling architecture PR (#25731) is still open, so these GGUFs need a patched build. Context is config.json's model_max_length (the card states none). Release date is the Hugging Face repo creation date.
Other places to get it
Checked Nov 14, 2026. We list only copies we opened and checked. Check file hashes for copies from other uploaders.
- ModelScope: Other uploaderSame account name as on Hugging Face; we found no link from the lab to it
Single GPUs that fit, cheapest first
| Single GPU that fits | VRAM | Speed estimate | Max context estimate | Price |
|---|---|---|---|---|
| Framework Desktop, AMD Ryzen AI Max+ PRO 495 (Strix Halo), 192 GB | 192 GB | ~27 tok/s | 1M (full) | low $6,799 |
Every device that runs it
Graphics cards, Macs and mini PCs from the wider hardware list behind My rig, smallest usable memory first. Not sure about yours? Check your exact rig.
Show all 4 devices that run it
| Device | Type | Memory | Speed estimate | Max context estimate | Price |
|---|---|---|---|---|---|
| Apple M3 Ultra 256GB | Mac | 256 GB (~192 usable) | ~67 tok/s | 1M (full) | No price yet |
| Apple M5 Ultra 256GB | Mac | 256 GB (~192 usable) | ~85 tok/s | 1M (full) | No price yet |
| Framework Desktop, AMD Ryzen AI Max+ PRO 495 (Strix Halo), 192 GB | Mini PC | 192 GB | ~27 tok/s | 1M (full) | low $6,799 |
| Apple M5 Ultra 512GB | Mac | 512 GB (~384 usable) | ~85 tok/s | 1M (full) | No price yet |
Multi-GPU: cards that need two or more
| Cards | Total VRAM | Speed estimate | Max context estimate |
|---|---|---|---|
| 4 × NVIDIA RTX A6000 | 192 GB | ~81 tok/s | 1M (full) |
Speed calculator
estimateEstimate = bandwidth ÷ GB read per token (the model's listed GGUF file: Q4_K_M for most models, otherwise the quant named in the list; scaled by active/total parameters for MoE), scaled by a per-platform efficiency and fixed per-token costs calibrated to 45 measured llama.cpp benchmarks (2025–26). Fit = file size + 1.5 GB headroom ≤ total VRAM (on Macs, the part of unified memory macOS lets the GPU use by default), and room for at least 4K tokens of context with an 8-bit KV cache. For a model with no KV cache data, only the file and headroom are checked, so a fit says nothing about context. Multi-GPU: VRAM adds, speed stays one card's (a layer split runs the cards in turn); max context uses the total. Speed is for short context. Max context = what is left after the file and a 1 GB runtime reserve ÷ KV cache bytes per token, default FP16 and 8-bit (q8_0) cache. How speed works · how context works