Model field guide
Qwen3-30B-A3B-Instruct-2507
26 of 44 hardware configurations have a supported memory fit at one or more recorded quantizations. Runtime compatibility still needs verification.
Total parameters30.53B
Active parameters3.30B
Recorded context limit262,144
Architecturemoe
Verified 2026-10-22 · license metadata: apache-2.0 · commercial use: yes (review the terms).
- Mixture of experts: weight memory uses total parameters. The separately sourced active-parameter figure is publisher-rounded and is not a reduction in required weight storage.
Actual files, not a size shortcut.
| Quantization | Total file size | Source |
|---|---|---|
| Q4_K_M | 17.282 GiB18,556,686,752 bytes; 1 file(s)Sum of exact file byte sizes in the repository listing at the recorded commit; every declared shard is present. Auxiliary vision projector and speculative-decoding files are excluded. Labels follow publisher filenames; mixed-precision expert tensors may remain in a differently labeled file, so the estimator uses measured bytes rather than assumed bits per weight. | Pinned file recordVerified 2026-10-22 |
| Q5_K_M | 20.234 GiB21,725,581,728 bytes; 1 file(s)Sum of exact file byte sizes in the repository listing at the recorded commit; every declared shard is present. Auxiliary vision projector and speculative-decoding files are excluded. Labels follow publisher filenames; mixed-precision expert tensors may remain in a differently labeled file, so the estimator uses measured bytes rather than assumed bits per weight. | Pinned file recordVerified 2026-10-22 |
| Q8_0 | 30.253 GiB32,483,932,576 bytes; 1 file(s)Sum of exact file byte sizes in the repository listing at the recorded commit; every declared shard is present. Auxiliary vision projector and speculative-decoding files are excluded. Labels follow publisher filenames; mixed-precision expert tensors may remain in a differently labeled file, so the estimator uses measured bytes rather than assumed bits per weight. | Pinned file recordVerified 2026-10-22 |
The hardware picture.
Q4_K_M: hardware comparison
| Hardware | Usable memory | Required memory | Fit estimate | Estimated generation |
|---|---|---|---|---|
| Mac mini M6 (12-core GPU) 16GB | 12.0 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M6 (12-core GPU) 24GB | 18.0 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M6 (12-core GPU) 32GB | 24.0 GiB | 18.66 GiB | fits | 33.9–67.8 estimated tokens/s |
| Mac mini M5 Pro (15-core CPU / 16-core GPU) 24GB | 18.0 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M5 Pro (15-core CPU / 16-core GPU) 48GB | 36.0 GiB | 18.66 GiB | fits | 61.2–122.5 estimated tokens/s |
| Mac mini M5 Pro (15-core CPU / 16-core GPU) 64GB | 48.0 GiB | 18.66 GiB | fits | 61.2–122.5 estimated tokens/s |
| Mac Studio M5 Max (32-core GPU) 36GB | 27.0 GiB | 18.66 GiB | fits | 91.7–183.5 estimated tokens/s |
| Mac Studio M5 Max (40-core GPU) 64GB | 48.0 GiB | 18.66 GiB | fits | 122.5–244.9 estimated tokens/s |
| Mac Studio M5 Max (40-core GPU) 128GB | 96.0 GiB | 18.66 GiB | fits | 122.5–244.9 estimated tokens/s |
| Mac Studio M5 Ultra (30-core CPU / 64-core GPU) 96GB | 72.0 GiB | 18.66 GiB | fits | 239.3–478.6 estimated tokens/s |
| Mac Studio M5 Ultra (36-core CPU / 80-core GPU) 256GB | 192.0 GiB | 18.66 GiB | fits | 239.3–478.6 estimated tokens/s |
| Mac Studio M5 Ultra (36-core CPU / 80-core GPU) 512GB | 384.0 GiB | 18.66 GiB | fits | 239.3–478.6 estimated tokens/s |
| MacBook Pro 14-inch M5 (10-core GPU) 24GB | 18.0 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| MacBook Pro 14-inch M5 Pro (20-core GPU) 48GB | 36.0 GiB | 18.66 GiB | fits | 61.2–122.5 estimated tokens/s |
| MacBook Pro 16-inch M5 Max (40-core GPU) 128GB | 96.0 GiB | 18.66 GiB | fits | 122.5–244.9 estimated tokens/s |
| Mac mini (2024) M4 (10-core GPU) 16GB | 12.0 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 (10-core GPU) 24GB | 18.0 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 (10-core GPU) 32GB | 24.0 GiB | 18.66 GiB | fits | 23.9–47.9 estimated tokens/s |
| Mac mini (2024) M4 Pro (16-core GPU) 24GB | 18.0 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 Pro (16-core GPU) 48GB | 36.0 GiB | 18.66 GiB | fits | 54.4–108.9 estimated tokens/s |
| Mac mini (2024) M4 Pro (16-core GPU) 64GB | 48.0 GiB | 18.66 GiB | fits | 54.4–108.9 estimated tokens/s |
| Mac Studio (2025) M4 Max (32-core GPU) 36GB | 27.0 GiB | 18.66 GiB | fits | 81.8–163.5 estimated tokens/s |
| Mac Studio (2025) M4 Max (40-core GPU) 64GB | 48.0 GiB | 18.66 GiB | fits | 108.9–217.8 estimated tokens/s |
| Mac Studio (2025) M4 Max (40-core GPU) 128GB | 96.0 GiB | 18.66 GiB | fits | 108.9–217.8 estimated tokens/s |
| Mac Studio (2025) M3 Ultra (28-core CPU / 60-core GPU) 96GB | 72.0 GiB | 18.66 GiB | fits | 163.3–326.7 estimated tokens/s |
| Mac Studio (2025) M3 Ultra (32-core CPU / 80-core GPU) 256GB | 192.0 GiB | 18.66 GiB | fits | 163.3–326.7 estimated tokens/s |
| NVIDIA GeForce RTX 3060 12GB | 10.8 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 3090 24GB | 21.6 GiB | 18.66 GiB | fits | 186.7–373.3 estimated tokens/s |
| NVIDIA GeForce RTX 4060 Ti 16GB | 14.4 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4070 12GB GDDR6X | 10.8 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4080 16GB | 14.4 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4090 24GB | 21.6 GiB | 18.66 GiB | fits | 201.0–402.1 estimated tokens/s |
| NVIDIA GeForce RTX 5060 Ti 16GB | 14.4 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5070 12GB | 10.8 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5070 Ti 16GB | 14.4 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5080 16GB | 14.4 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5090 32GB | 28.8 GiB | 18.66 GiB | fits | 357.4–714.8 estimated tokens/s |
| NVIDIA RTX PRO 6000 Blackwell Workstation Edition 96GB | 86.4 GiB | 18.66 GiB | fits | 357.4–714.8 estimated tokens/s |
| NVIDIA RTX 6000 Ada Generation 48GB | 43.2 GiB | 18.66 GiB | fits | 191.5–382.9 estimated tokens/s |
| Intel Arc B580 12GB | 10.8 GiB | 18.66 GiB | doesn't fit | Estimated speed unavailable |
| AMD Ryzen AI Halo, Ryzen AI Max+ 395, 128GB | 89.6 GiB | 18.66 GiB | fits | 51.1–102.1 estimated tokens/s |
| NVIDIA DGX Spark 128GB / 4TB | 96.0 GiB | 18.66 GiB | fits | 54.4–108.9 estimated tokens/s |
| 2 × GeForce RTX 3090 24GB (conditional 48GB aggregate) | 40.8 GiB | 18.66 GiB | Needs verification | Estimated speed unavailable |
| 2 × GeForce RTX 5090 32GB (conditional 64GB aggregate) | 54.4 GiB | 18.66 GiB | Needs verification | Estimated speed unavailable |
Q5_K_M: hardware comparison
| Hardware | Usable memory | Required memory | Fit estimate | Estimated generation |
|---|---|---|---|---|
| Mac mini M6 (12-core GPU) 16GB | 12.0 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M6 (12-core GPU) 24GB | 18.0 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M6 (12-core GPU) 32GB | 24.0 GiB | 21.61 GiB | borderline | 29.0–57.9 estimated tokens/s |
| Mac mini M5 Pro (15-core CPU / 16-core GPU) 24GB | 18.0 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M5 Pro (15-core CPU / 16-core GPU) 48GB | 36.0 GiB | 21.61 GiB | fits | 52.3–104.6 estimated tokens/s |
| Mac mini M5 Pro (15-core CPU / 16-core GPU) 64GB | 48.0 GiB | 21.61 GiB | fits | 52.3–104.6 estimated tokens/s |
| Mac Studio M5 Max (32-core GPU) 36GB | 27.0 GiB | 21.61 GiB | fits | 78.4–156.7 estimated tokens/s |
| Mac Studio M5 Max (40-core GPU) 64GB | 48.0 GiB | 21.61 GiB | fits | 104.6–209.2 estimated tokens/s |
| Mac Studio M5 Max (40-core GPU) 128GB | 96.0 GiB | 21.61 GiB | fits | 104.6–209.2 estimated tokens/s |
| Mac Studio M5 Ultra (30-core CPU / 64-core GPU) 96GB | 72.0 GiB | 21.61 GiB | fits | 204.4–408.8 estimated tokens/s |
| Mac Studio M5 Ultra (36-core CPU / 80-core GPU) 256GB | 192.0 GiB | 21.61 GiB | fits | 204.4–408.8 estimated tokens/s |
| Mac Studio M5 Ultra (36-core CPU / 80-core GPU) 512GB | 384.0 GiB | 21.61 GiB | fits | 204.4–408.8 estimated tokens/s |
| MacBook Pro 14-inch M5 (10-core GPU) 24GB | 18.0 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| MacBook Pro 14-inch M5 Pro (20-core GPU) 48GB | 36.0 GiB | 21.61 GiB | fits | 52.3–104.6 estimated tokens/s |
| MacBook Pro 16-inch M5 Max (40-core GPU) 128GB | 96.0 GiB | 21.61 GiB | fits | 104.6–209.2 estimated tokens/s |
| Mac mini (2024) M4 (10-core GPU) 16GB | 12.0 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 (10-core GPU) 24GB | 18.0 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 (10-core GPU) 32GB | 24.0 GiB | 21.61 GiB | borderline | 20.4–40.9 estimated tokens/s |
| Mac mini (2024) M4 Pro (16-core GPU) 24GB | 18.0 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 Pro (16-core GPU) 48GB | 36.0 GiB | 21.61 GiB | fits | 46.5–93.0 estimated tokens/s |
| Mac mini (2024) M4 Pro (16-core GPU) 64GB | 48.0 GiB | 21.61 GiB | fits | 46.5–93.0 estimated tokens/s |
| Mac Studio (2025) M4 Max (32-core GPU) 36GB | 27.0 GiB | 21.61 GiB | fits | 69.8–139.7 estimated tokens/s |
| Mac Studio (2025) M4 Max (40-core GPU) 64GB | 48.0 GiB | 21.61 GiB | fits | 93.0–186.0 estimated tokens/s |
| Mac Studio (2025) M4 Max (40-core GPU) 128GB | 96.0 GiB | 21.61 GiB | fits | 93.0–186.0 estimated tokens/s |
| Mac Studio (2025) M3 Ultra (28-core CPU / 60-core GPU) 96GB | 72.0 GiB | 21.61 GiB | fits | 139.5–279.0 estimated tokens/s |
| Mac Studio (2025) M3 Ultra (32-core CPU / 80-core GPU) 256GB | 192.0 GiB | 21.61 GiB | fits | 139.5–279.0 estimated tokens/s |
| NVIDIA GeForce RTX 3060 12GB | 10.8 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 3090 24GB | 21.6 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4060 Ti 16GB | 14.4 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4070 12GB GDDR6X | 10.8 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4080 16GB | 14.4 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4090 24GB | 21.6 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5060 Ti 16GB | 14.4 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5070 12GB | 10.8 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5070 Ti 16GB | 14.4 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5080 16GB | 14.4 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5090 32GB | 28.8 GiB | 21.61 GiB | fits | 305.3–610.5 estimated tokens/s |
| NVIDIA RTX PRO 6000 Blackwell Workstation Edition 96GB | 86.4 GiB | 21.61 GiB | fits | 305.3–610.5 estimated tokens/s |
| NVIDIA RTX 6000 Ada Generation 48GB | 43.2 GiB | 21.61 GiB | fits | 163.5–327.1 estimated tokens/s |
| Intel Arc B580 12GB | 10.8 GiB | 21.61 GiB | doesn't fit | Estimated speed unavailable |
| AMD Ryzen AI Halo, Ryzen AI Max+ 395, 128GB | 89.6 GiB | 21.61 GiB | fits | 43.6–87.2 estimated tokens/s |
| NVIDIA DGX Spark 128GB / 4TB | 96.0 GiB | 21.61 GiB | fits | 46.5–93.0 estimated tokens/s |
| 2 × GeForce RTX 3090 24GB (conditional 48GB aggregate) | 40.8 GiB | 21.61 GiB | Needs verification | Estimated speed unavailable |
| 2 × GeForce RTX 5090 32GB (conditional 64GB aggregate) | 54.4 GiB | 21.61 GiB | Needs verification | Estimated speed unavailable |
Q8_0: hardware comparison
| Hardware | Usable memory | Required memory | Fit estimate | Estimated generation |
|---|---|---|---|---|
| Mac mini M6 (12-core GPU) 16GB | 12.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M6 (12-core GPU) 24GB | 18.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M6 (12-core GPU) 32GB | 24.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M5 Pro (15-core CPU / 16-core GPU) 24GB | 18.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini M5 Pro (15-core CPU / 16-core GPU) 48GB | 36.0 GiB | 31.63 GiB | fits | 35.0–70.0 estimated tokens/s |
| Mac mini M5 Pro (15-core CPU / 16-core GPU) 64GB | 48.0 GiB | 31.63 GiB | fits | 35.0–70.0 estimated tokens/s |
| Mac Studio M5 Max (32-core GPU) 36GB | 27.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac Studio M5 Max (40-core GPU) 64GB | 48.0 GiB | 31.63 GiB | fits | 70.0–139.9 estimated tokens/s |
| Mac Studio M5 Max (40-core GPU) 128GB | 96.0 GiB | 31.63 GiB | fits | 70.0–139.9 estimated tokens/s |
| Mac Studio M5 Ultra (30-core CPU / 64-core GPU) 96GB | 72.0 GiB | 31.63 GiB | fits | 136.7–273.4 estimated tokens/s |
| Mac Studio M5 Ultra (36-core CPU / 80-core GPU) 256GB | 192.0 GiB | 31.63 GiB | fits | 136.7–273.4 estimated tokens/s |
| Mac Studio M5 Ultra (36-core CPU / 80-core GPU) 512GB | 384.0 GiB | 31.63 GiB | fits | 136.7–273.4 estimated tokens/s |
| MacBook Pro 14-inch M5 (10-core GPU) 24GB | 18.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| MacBook Pro 14-inch M5 Pro (20-core GPU) 48GB | 36.0 GiB | 31.63 GiB | fits | 35.0–70.0 estimated tokens/s |
| MacBook Pro 16-inch M5 Max (40-core GPU) 128GB | 96.0 GiB | 31.63 GiB | fits | 70.0–139.9 estimated tokens/s |
| Mac mini (2024) M4 (10-core GPU) 16GB | 12.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 (10-core GPU) 24GB | 18.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 (10-core GPU) 32GB | 24.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 Pro (16-core GPU) 24GB | 18.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac mini (2024) M4 Pro (16-core GPU) 48GB | 36.0 GiB | 31.63 GiB | fits | 31.1–62.2 estimated tokens/s |
| Mac mini (2024) M4 Pro (16-core GPU) 64GB | 48.0 GiB | 31.63 GiB | fits | 31.1–62.2 estimated tokens/s |
| Mac Studio (2025) M4 Max (32-core GPU) 36GB | 27.0 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| Mac Studio (2025) M4 Max (40-core GPU) 64GB | 48.0 GiB | 31.63 GiB | fits | 62.2–124.4 estimated tokens/s |
| Mac Studio (2025) M4 Max (40-core GPU) 128GB | 96.0 GiB | 31.63 GiB | fits | 62.2–124.4 estimated tokens/s |
| Mac Studio (2025) M3 Ultra (28-core CPU / 60-core GPU) 96GB | 72.0 GiB | 31.63 GiB | fits | 93.3–186.6 estimated tokens/s |
| Mac Studio (2025) M3 Ultra (32-core CPU / 80-core GPU) 256GB | 192.0 GiB | 31.63 GiB | fits | 93.3–186.6 estimated tokens/s |
| NVIDIA GeForce RTX 3060 12GB | 10.8 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 3090 24GB | 21.6 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4060 Ti 16GB | 14.4 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4070 12GB GDDR6X | 10.8 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4080 16GB | 14.4 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 4090 24GB | 21.6 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5060 Ti 16GB | 14.4 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5070 12GB | 10.8 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5070 Ti 16GB | 14.4 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5080 16GB | 14.4 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA GeForce RTX 5090 32GB | 28.8 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| NVIDIA RTX PRO 6000 Blackwell Workstation Edition 96GB | 86.4 GiB | 31.63 GiB | fits | 204.2–408.3 estimated tokens/s |
| NVIDIA RTX 6000 Ada Generation 48GB | 43.2 GiB | 31.63 GiB | fits | 109.4–218.7 estimated tokens/s |
| Intel Arc B580 12GB | 10.8 GiB | 31.63 GiB | doesn't fit | Estimated speed unavailable |
| AMD Ryzen AI Halo, Ryzen AI Max+ 395, 128GB | 89.6 GiB | 31.63 GiB | fits | 29.2–58.3 estimated tokens/s |
| NVIDIA DGX Spark 128GB / 4TB | 96.0 GiB | 31.63 GiB | fits | 31.1–62.2 estimated tokens/s |
| 2 × GeForce RTX 3090 24GB (conditional 48GB aggregate) | 40.8 GiB | 31.63 GiB | Needs verification | Estimated speed unavailable |
| 2 × GeForce RTX 5090 32GB (conditional 64GB aggregate) | 54.4 GiB | 31.63 GiB | Needs verification | Estimated speed unavailable |
A “fits” result checks estimated memory only. “Needs verification” preserves missing cache, vision, or placement information. All speed figures are estimated from the published bandwidth formula.
Field-level sources and verification dates
- id: Official source · verified 2026-10-22. Lowercase URL slug derived from the upstream repository ID; baseModel retains the exact Hugging Face identifier.
- name: Official source · verified 2026-10-22
- baseModel: Official source · verified 2026-10-22
- downloads: Official source · verified 2026-10-22
- tasks: Official source · verified 2026-10-22. Chat denotes a text-generating model, including base models that may require a chat template or instruction tuning. Code label is inferred from the repository model-family name. Vision uses pipeline metadata or vision_config.
- license: Official source · verified 2026-10-22
- commercialUse: Official source · verified 2026-10-22. Official upstream organization metadata identifies a permissive standard license. Commercial use is permitted subject to the license terms, notices, and applicable restrictions; this is a metadata-based classification, not an individualized legal approval.
- license_url: Official source · verified 2026-10-22. Pinned license file confirmed in repository listing.
- architecture: Official source · verified 2026-10-22. Classification from model_type and expert/state-space configuration fields (qwen3_moe); hybrid classification takes precedence over expert count.
- parametersB: Official source · verified 2026-10-22. Exact safetensors parameter total reported by the Hugging Face API, divided by 1e9.
- activeParametersB: Official source · verified 2026-10-22. Official pinned model card explicitly states 3.3B activated parameters. Publisher-rounded figure in billions, not an exact integer tensor count. Transcribed from the model-specific statement, never inferred from the model name. Total parameter memory remains separate.
- contextLength: Official source · verified 2026-10-22. Configured maximum token positions; not a guarantee of runtime support or quality at that length. No speculative RoPE scaling applied.
- kvCache: Official source · verified 2026-10-22. Standard attention KV shape from num_hidden_layers, num_key_value_heads, and head_dim or hidden_size / num_attention_heads; FP16 cache bytes are a separate estimator choice.
- licenseStandard: Official source · verified 2026-10-22. Official standard license text supporting the permissive-license classification; model applicability is identified separately by upstream metadata.