AI Hardware & Model Index
Every machine that runs AI locally, with configurations, prices, memory, bandwidth and compute — and how each model fits and how fast it runs. Below it, the keystone of hosted and open models.
67
Configurations
23
Vendors
66
Models
2026-11-05
Last refreshed
Specs come from vendor pages; speeds and fit are labelled estimates. Prices are dated snapshots — confirm with the vendor before buying. No affiliate links.
66 models · click a row for pricing, downloads and the machines that run it.
| Model | Weights | Params | Context | Modalities | Hosted $/1M in · out | Local file | Fits on | Released |
|---|---|---|---|---|---|---|---|---|
GPT-6.1 Terra OpenAI | Hosted | — | 1,050K | text, image | $1 · $5 | — | — | 2026-10-26 |
Nemotron 3.5 Bolt E4B NVIDIA | Open | 8B 4B active | 66K | text | — | Q4_K_M · 6.2 GB | 65 machines | 2026-10-23 |
Flint Recursive | Open | 40B 2.5B active | 262K | text | — | Q4_K_M · 25.4 GB | 47 machines | 2026-10-22 |
DeepSeek-V4.1-Flash-1017 DeepSeek | Open | 748B 16B active | 1,049K | text, image | $0.15 · $0.6 | UD-Q4_K_XL · 501 GB | 0 machines | 2026-10-17 |
Nemotron 3.5 Bolt 74B-A9B NVIDIA | Open | 74B 9B active | 262K | text, image, video | — | IQ2_XXS experts + NVFP4 Mamba + Q8_0 embeddings (QAT) · 22 GB | 47 machines | 2026-10-15 |
Nemotron 3.5 Super VL 120B-A12B NVIDIA | Open | 120B 12B active | 1,049K | text, image, video | — | Q4_K_M · 90 GB | 33 machines | 2026-10-15 |
Muse Spark Open 816B-A42B Meta | Open | 816B 42B active | 1,049K | text, image, audio, video | — | UD-Q4_K_XL · 506 GB | 0 machines | 2026-10-15 |
Gemini 4 Argon Google | Hosted | — | 1,049K | text, image, audio, video | $2 · $10 | — | — | 2026-10-13 |
Claude Haiku 5.5 Anthropic | Hosted | — | 1,000K | text, image | $0.1 · $0.5 | — | — | 2026-10-07 |
Mistral Large 4 (1T-A52B) Mistral AI | Open | 1,050B 52B active | 1,049K | text, image | $0.7 · $2.1 | Q4_K_M · 635 GB | 0 machines | 2026-10 |
GPT-6.1 Sol OpenAI | Hosted | — | 1,050K | text, image | $2 · $10 | — | — | 2026-09-29 |
Claude Sonnet 5.5 Anthropic | Hosted | — | 1,000K | text, image | $2 · $10 | — | — | 2026-09-28 |
GPT-6 Luna OpenAI | Hosted | — | 1,050K | text, image | $0.1 · $0.5 | — | — | 2026-09-22 |
Claude Opus 5.5 Anthropic | Hosted | — | 1,000K | text, image | $4 · $20 | — | — | 2026-09-22 |
Grok 4.7 xAI | Hosted | — | 500K | text, image | $2 · $6 | — | — | 2026-09-21 |
GPT-6 Astra OpenAI | Hosted | — | 1,050K | text, image | $10 · $50 | — | — | 2026-09-03 |
Gemini 3.8 Flash Google | Hosted | — | 1,049K | text, image, audio, video | $0.75 · $3.75 | — | — | 2026-09-02 |
Claude Fable 5.1 Anthropic | Hosted | — | 1,000K | text, image | $10 · $50 | — | — | 2026-09-01 |
Granite 4.2 3B IBM | Open | 4B | 131K | text | — | Q4_K_M · 2.2 GB | 66 machines | 2026-08-25 |
Granite 4.2 8B IBM | Open | 9B | 131K | text | — | Q4_K_M · 5.3 GB | 65 machines | 2026-08-25 |
Granite 4.2 30B IBM | Open | 29B | 131K | text | — | Q4_K_M · 17.7 GB | 48 machines | 2026-08-25 |
Nemotron 3.5 Lightning 30B-A3B NVIDIA | Open | 30B 3B active | 1,000K | text | — | Q4_0 · 18.9 GB | 48 machines | 2026-08-11 |
Qwen3.8-27B Alibaba | Open | 27B | 262K | text, image, video | — | Q4_K_M · 19 GB | 48 machines | 2026-08 |
Qwen3.8-Flash-Next Alibaba | Open | 180B 6B active | 262K | text, image, video | — | UD-Q4_K_XL · 111.3 GB | 9 machines | 2026-08 |
Qwen3.8-2.4T-A95B Alibaba | Open | 2,400B 95B active | 262K | text | — | UD-IQ4_XS · 1,310 GB | 0 machines | 2026-08 |
Muse Glimmer 30B Meta | Open | 30B | 131K | text, image | — | Q4_K_M · 16.8 GB | 61 machines | 2026-08 |
Gemini 3.5 Flash-Lite Google | Hosted | — | 1,049K | text, image, audio, video | $0.3 · $2.5 | — | — | 2026-07-21 |
Gemini 3.6 Flash Google | Hosted | — | — | — | $0.75 · $3.75 | — | — | 2026-07-21 |
Command A+ (05-2026) Cohere | Open | 218B 25B active | 128K | text, image | — | Q4_K_M · 135.1 GB | 4 machines | 2026-05-20 |
Mistral Medium 3.5 128B Mistral AI | Open | 128B | 262K | — | — | Q4_K_M · 74.9 GB | 37 machines | 2026-04-28 |
Gemma 4 26B A4B Google | Open | 25B 3.8B active | 262K | text, image | — | Q4_K_M · 18 GB | 61 machines | 2026-04-02 |
Gemma 4 31B Google | Open | 31B | 262K | text, image | — | Q4_K_M · 20 GB | 48 machines | 2026-04-02 |
Qwen3.6-35B-A3B Alibaba | Open | 35B 3B active | 262K | text, image, video | — | Q4_K_M · 20.4 GB | 48 machines | 2026-04 |
Phi-4-reasoning-vision-15B Microsoft | Open | 15B | 16K | text, image | — | Q4_K_M (community, text-only) · 9.1 GB | 64 machines | 2026-03-04 |
GPT-5.3 Codex OpenAI | Hosted | — | — | — | $1.75 · $14 | — | — | 2026-02-24 |
Amazon Nova 2 Lite Amazon | Hosted | — | 1,000K | text, image, video | — | — | — | 2025-12-02 |
gpt-oss-20b OpenAI | Open | 21B 3.6B active | 131K | text | — | MXFP4 (MoE weights) + BF16 (original checkpoint) · 13.8 GB | 61 machines | 2025-08 |
gpt-oss-120b OpenAI | Open | 117B 5.1B active | 131K | text | — | MXFP4 (MoE weights) · 65 GB | 37 machines | 2025-08 |
Llama 4 Maverick 17B-128E Meta | Open | 400B 17B active | 1,000K | text, image | — | Q4_K_M · 245 GB | 1 machines | 2025-04-05 |
Llama 4 Scout 17B-16E Meta | Open | 109B 17B active | 10,000K | text, image | — | Q4_K_M · 67 GB | 37 machines | 2025-04-05 |
LFM2.5-8B-A1B Liquid AI | Open | 8B 1.5B active | 128K | text | — | Q4_K_M · 5.2 GB | 65 machines | — |
LFM2.5-24B-A2B Liquid AI | Open | 24B 2.5B active | 131K | text | — | Q4_K_M · 14.5 GB | 61 machines | — |
MiniMax-M3 MiniMax | Open | 427B 23B active | 1,049K | text, image, video | $0.3 · $1.2 | UD-Q4_K_M · 265 GB | 1 machines | — |
GLM-5.3-Flash Zhipu AI (Z.ai) | Open | 320B 18B active | 1,049K | text, image | $0.15 · $0.5 | UD-Q4_K_XL · 200 GB | 4 machines | — |
GLM-5.3 Zhipu AI (Z.ai) | Open | 753B | 1,049K | text | $1.4 · $4.4 | UD-Q4_K_XL · 467 GB | 0 machines | — |
Kimi K3.1 Pro Moonshot AI | Open | 2,800B 104B active | 1,049K | text, image, video | — | MXFP4 weights / MXFP8 activations (native) · 1,560 GB | 0 machines | — |
Kimi K3 Moonshot AI | Open | 2,800B 104B active | 1,049K | text, image, video | — | UD-Q4_K_XL · 1,510 GB | 0 machines | — |
Qwen3-Coder-Next Alibaba | Open | 80B 3B active | 262K | text | — | Q4_K_M · 48.4 GB | 39 machines | — |
DeepSeek-V4-Pro DeepSeek | Open | 1,600B 49B active | 1,049K | text | $0.66 · $1.98 | NVFP4 · 913 GB | 0 machines | — |
Ministral 3 3B Mistral AI | Open | 4B | 262K | text, image | $0.1 · $0.1 | Q4_K_M · 3 GB | 66 machines | — |
Ministral 3 8B Mistral AI | Open | 9B | 262K | text, image | $0.15 · $0.15 | Q4_K_M · 5.2 GB | 65 machines | — |
Ministral 3 14B Mistral AI | Open | 14B | 262K | text, image | $0.2 · $0.2 | Q4_K_M · 8.2 GB | 64 machines | — |
Mistral Small 4 119B-2603 Mistral AI | Open | 119B 6.5B active | 262K | text, image | $0.15 · $0.6 | UD-Q4_K_M · 73.8 GB | 37 machines | — |
Gemma 4 E2B Google | Open | 5B 2.3B active | 131K | text, image, audio | — | Q4_K_M · 4.6 GB | 66 machines | — |
Gemma 4 E4B Google | Open | 8B 4.5B active | 131K | text, image, audio | — | Q4_K_M · 6.6 GB | 65 machines | — |
Gemma 4 12B Unified Google | Open | 12B | 262K | text, image, audio | — | Q4_K_M · 8 GB | 64 machines | — |
MiniMax-M3.1-Flash MiniMax | Hosted | — | — | — | $0.15 · $0.6 | — | — | — |
Qwen3.8-Flash Alibaba | Hosted | — | — | — | $0.15 · $0.47 | — | — | — |
Qwen3.8-Max Alibaba | Hosted | — | 1,000K | text, image | $2 · $6 | — | — | — |
Gemini 3.1 Pro (preview) Google | Hosted | — | — | — | $2 · $12 | — | — | — |
Gemini 3.1 Flash-Lite Google | Hosted | — | — | — | $0.25 · $1.5 | — | — | — |
Grok 4.20 (0309 reasoning) xAI | Hosted | — | 1,000K | text, image | $1.25 · $2.5 | — | — | — |
Muse Wonder 1.0 Meta | Hosted | — | 1,049K | — | $4 · $20 | — | — | — |
Grok 4.3 xAI | Hosted | — | 1,000K | — | $1.25 · $2.5 | — | — | — |
Grok Build 0.1 xAI | Hosted | — | — | — | $1 · $2 | — | — | — |
Muse Spark 1.3 Meta | Hosted | — | 1,049K | text, image, video | $0.1 · $0.2 | — | — | — |