Skip to content
English

Switch theme

Discover▾

Not sure where to start? This week’s hottest software, models and stacks in three minutes.

AI Apps▾

AI software people use: from assistants like ChatGPT to tools for coding, video and workflows, picked by purpose.

For Agents▾

Software for AI: tools that let your agent (Claude Code, Codex, Cursor…) browse, read docs, run code and remember.

Models▾

Choose models, estimate cost, self-host: prices, rankings, benchmarks, free tiers and compute.

Solutions▾

A goal or an error message? Ready-made stacks, steps and fixes by scenario.

Learn▾

A systematic course from zero to the frontier, with an algorithm library, must-read papers and tutorials.

Frontier▾

Daily AI news, weekly highlights, the latest papers and how AI is changing every discipline.

// Open-model VRAM table · family

IBM Granite: local deployment & VRAM

67 open-weight models from IBM (147 official repos incl. quantised builds). At 4-bit they range from 0.8 GB (granite-4.0-h-350m) to 22.7 GB (granite-switch-4.1-30b).

Filter by my hardware → VRAM calculator

official orgs: ibm-granitelatest 2026-10-14memory = estimate

01Text LLM (32)

ModelParamsContextBF164-bitSmallest GPU (4-bit)Mac (4-bit)LicenceReleased
granite-switch-4.1-8b8.78B 128K 21.2 GB7.8 GB 12GB GPUMac 16GB Commercial OK2026-10-14
granite-switch-4.1-30b30B 128K 68.7 GB22.7 GB RTX 5090 32GBMac 64GB Commercial OK2026-10-14
ensemble-forecasting-with-ibm-granite-time-series— — —— —— Licence not stated2026-10-01
granite-4.2-3b3.66B 128K 9.2 GB3.6 GB 8GB GPUMac 16GB Commercial OK2026-08-07
granite-4.2-8b8.79B 128K 21.2 GB7.7 GB 12GB GPUMac 16GB Commercial OK2026-08-07
granite-4.2-30b29.3B 128K 67.1 GB22.2 GB RTX 5090 32GBMac 64GB Commercial OK2026-08-07
granite-4.1-3b3.4B 128K 8.7 GB3.4 GB 8GB GPUMac 16GB Commercial OK2026-04-06
granite-4.1-8b8.79B 128K 21.2 GB7.7 GB 12GB GPUMac 16GB Commercial OK2026-04-06
granite-4.1-30b28.9B 128K 66.2 GB21.9 GB RTX 5090 32GBMac 64GB Commercial OK2026-04-06
granitelib-core-r1.0— — —— —— Commercial OK2025-12-10
granite-4.0-350m352M 32K 1.5 GB1.0 GB 8GB GPUMac 16GB Commercial OK2025-10-07
granite-4.0-1b1.63B 128K 4.8 GB2.3 GB 8GB GPUMac 16GB Commercial OK2025-10-07
granite-4.0-h-350m340M 32K 1.3 GB0.8 GB 8GB GPUMac 16GB Commercial OK2025-10-07
granite-4.0-h-1b1.46B 128K 3.8 GB1.5 GB 8GB GPUMac 16GB Commercial OK2025-10-07
granite-4.0-h-micro3.19B 128K 7.6 GB2.7 GB 8GB GPUMac 16GB Commercial OK2025-09-16
granite-4.0-h-small32.2B 128K 71.5 GB22.1 GB RTX 5090 32GBMac 64GB Commercial OK2025-09-16
granite-4.0-h-tiny6.94B 128K 15.8 GB5.2 GB 8GB GPUMac 16GB Commercial OK2025-09-16
granite-4.0-micro3.4B 128K 8.7 GB3.4 GB 8GB GPUMac 16GB Commercial OK2025-09-16
granite-4.0-tiny-preview6.67B 128K 15.2 GB5.0 GB 8GB GPUMac 16GB Commercial OK2025-04-30
granite-3.3-2b-instruct2.53B 128K 6.7 GB2.9 GB 8GB GPUMac 16GB Commercial OK2025-04-09
granite-3.3-8b-instruct8.17B 128K 19.8 GB7.3 GB 12GB GPUMac 16GB Commercial OK2025-04-09
granite-3.2-2b-instruct2.53B 128K 6.7 GB2.9 GB 8GB GPUMac 16GB Commercial OK2025-02-17
granite-3.2-8b-instruct8.17B 128K 19.8 GB7.3 GB 12GB GPUMac 16GB Commercial OK2025-02-17
granite-3.1-2b-instruct2.53B 128K 6.7 GB2.9 GB 8GB GPUMac 16GB Commercial OK2024-12-06
granite-3.1-8b-instruct8.17B 128K 19.8 GB7.3 GB 12GB GPUMac 16GB Commercial OK2024-12-06
granite-3.1-1b-a400m-instruct1.33B 128K 3.8 GB1.8 GB 8GB GPUMac 16GB Commercial OK2024-12-06
granite-3.1-3b-a800m-instruct3.3B 128K 8.3 GB3.2 GB 8GB GPUMac 16GB Commercial OK2024-12-06
granite-3.0-1b-a400m-instruct1.33B / 400M 4K 3.8 GB1.8 GB 8GB GPUMac 16GB Commercial OK2024-10-03
granite-3.0-3b-a800m-instruct3.37B / 800M 4K 8.5 GB3.3 GB 8GB GPUMac 16GB Commercial OK2024-10-03
granite-3.0-2b-instruct2.63B 4K 7.0 GB2.9 GB 8GB GPUMac 16GB Commercial OK2024-10-02
granite-3.0-8b-instruct8.17B 4K 19.8 GB7.3 GB 12GB GPUMac 16GB Commercial OK2024-10-02
granite-8b-code-instruct-128k ↗8.05B 125K 19.4 GB7.1 GB 8GB GPUMac 16GB Commercial OK2024-07-12

02Vision-language (5)

ModelParamsContextBF164-bitSmallest GPU (4-bit)Mac (4-bit)LicenceReleased
granite-vision-4.1-4b4B 128K 10.0 GB3.8 GB 8GB GPUMac 16GB Commercial OK2026-04-16
granite-4.0-3b-vision4B 128K 10.0 GB3.8 GB 8GB GPUMac 16GB Commercial OK2026-03-03
granite-vision-3.3-2b2.98B 128K 7.7 GB3.2 GB 8GB GPUMac 16GB Commercial OK2025-06-03
granite-docling-258M258M 8K 1.3 GB0.9 GB 8GB GPUMac 16GB Commercial OK2025-05-19
granite-vision-3.2-2b2.98B 128K 7.7 GB3.2 GB 8GB GPUMac 16GB Commercial OK2025-02-17

03Embedding (10)

ModelParamsBF164-bitSmallest GPU (4-bit)Mac (4-bit)LicenceReleased
granite-embedding-97m-multilingual-r297M 0.7 GB0.6 GB 8GB GPUMac 16GB Commercial OK2026-04-20
granite-embedding-311m-multilingual-r2312M 1.2 GB0.7 GB 8GB GPUMac 16GB Commercial OK2026-04-20
granite-speech-4.1-2b-nar2.25B 5.5 GB2.0 GB 8GB GPUMac 16GB Commercial OK2026-03-10
granite-embedding-english-r2149M 0.8 GB0.6 GB 8GB GPUMac 16GB Commercial OK2025-07-17
granite-embedding-small-english-r248M 0.6 GB0.5 GB 8GB GPUMac 16GB Commercial OK2025-07-17
granite-embedding-30m-sparse30M 0.6 GB0.5 GB 8GB GPUMac 16GB Commercial OK2025-02-17
granite-embedding-30m-english30M 0.6 GB0.5 GB 8GB GPUMac 16GB Commercial OK2024-12-04
granite-embedding-125m-english125M 0.8 GB0.6 GB 8GB GPUMac 16GB Commercial OK2024-12-04
granite-embedding-107m-multilingual107M 0.7 GB0.6 GB 8GB GPUMac 16GB Commercial OK2024-12-04
granite-embedding-278m-multilingual278M 1.1 GB0.7 GB 8GB GPUMac 16GB Commercial OK2024-12-04

04Reranker (1)

ModelParamsBF164-bitSmallest GPU (4-bit)Mac (4-bit)LicenceReleased
granite-embedding-reranker-english-r2150M 0.8 GB0.6 GB 8GB GPUMac 16GB Commercial OK2025-08-04

05Speech-to-text (9)

ModelParamsBF164-bitSmallest GPU (4-bit)Mac (4-bit)LicenceReleased
granite-speech-5.0-2b2.31B 5.6 GB2.0 GB 8GB GPUMac 16GB Commercial OK2026-10-07
granite-speech-5.0-470m-turboctc-nc473M 1.5 GB0.8 GB 8GB GPUMac 16GB Non-commercial2026-08-17
granite-speech-5.0-470m-turboctc473M 1.5 GB0.8 GB 8GB GPUMac 16GB Commercial OK2026-08-04
granite-speech-4.1-2b2.31B 5.6 GB2.0 GB 8GB GPUMac 16GB Commercial OK2026-04-16
granite-speech-4.1-2b-plus2.11B 5.1 GB1.9 GB 8GB GPUMac 16GB Commercial OK2026-04-16
granite-4.0-1b-speech2.31B 5.6 GB2.0 GB 8GB GPUMac 16GB Commercial OK2026-02-27
granite-speech-3.3-2b3.01B 7.1 GB2.5 GB 8GB GPUMac 16GB Commercial OK2025-04-28
granite-speech-3.3-8b8.65B 19.5 GB6.3 GB 8GB GPUMac 16GB Commercial OK2025-04-14
granite-speech-3.2-8b8.48B 19.2 GB6.2 GB 8GB GPUMac 16GB Commercial OK2025-03-26

06Safety / guard (10)

ModelParamsContextBF164-bitSmallest GPU (4-bit)Mac (4-bit)LicenceReleased
granite-guardian-4.1-8b8.38B 128K 20.3 GB7.4 GB 12GB GPUMac 16GB Commercial OK2026-04-16
granitelib-guardian-r1.0 ↗— — —— —— Commercial OK2026-02-02
granite-guardian-3.3-8b8.17B 128K 19.8 GB7.3 GB 12GB GPUMac 16GB Commercial OK2025-06-03
granite-guardian-3.2-3b-a800m ↗3.3B 128K 8.3 GB3.2 GB 8GB GPUMac 16GB Commercial OK2025-02-03
granite-guardian-3.2-5b ↗5.78B 128K 14.2 GB5.3 GB 8GB GPUMac 16GB Commercial OK2025-01-23
granite-guardian-3.1-2b ↗2.53B 128K 6.7 GB2.9 GB 8GB GPUMac 16GB Commercial OK2024-12-17
granite-guardian-3.1-8b ↗8.17B 128K 19.8 GB7.3 GB 12GB GPUMac 16GB Commercial OK2024-12-17
granite-guardian-3.0-2b2.53B 8K 6.7 GB2.9 GB 8GB GPUMac 16GB Commercial OK2024-10-15
granite-guardian-hap-38m ↗38M 1K 0.7 GB0.6 GB 8GB GPUMac 16GB Commercial OK2024-09-05
granite-guardian-hap-125m ↗125M 1K 1.1 GB0.9 GB 8GB GPUMac 16GB Commercial OK2024-09-05

07Other families

Qwen · NVIDIA Nemotron · Gemma · Mistral / Magistral / Devstral · GLM / CogVideoX (Z.ai) · DeepSeek · Ai2 OLMo / Molmo · Ant Ling / Ring / Ming · Liquid LFM · MiniCPM / VoxCPM · Tencent Hunyuan / Hy · Phi · Llama · Meituan LongCat · Jina embeddings & rerankers · Wan · Kimi · Cohere Command / Aya · Xiaomi MiMo · FLUX · Microsoft VibeVoice / Fara / Harrier · InternLM / Intern-S · StepFun Step · MiniMax · Baidu ERNIE · NVIDIA Parakeet / Canary speech · Stable Diffusion / Stable Audio · Yi (01.AI) · ByteDance Seed · SmolLM / SmolVLM · BGE embeddings & rerankers · LTX Video · FunAudioLLM speech · gpt-oss · poolside Laguna · Whisper · Baichuan · Thinking Machines Inkling · xAI Grok · rednote dots

Totals include 10% + 0.5GB overhead and one 8K-token KV cache for text models. Estimates only. Method →

Data updated: Self-host open models(UTC+8)