# Qwen3 8B datasheet

> A small dense model with switchable thinking. The pick for 8 to 12 GB cards when you want Apache-2.0 terms.

Canonical: https://thelocalbench.com/models/qwen3-8b · Data as of 2026-10-12 · The Local Bench (thelocalbench.com), data as of 2026-10-12, CC BY 4.0

Flags: [verified] checked against primary source · [sourced] cited · [estimate] formula · [sample] placeholder

## Specifications

| Field | Value | Flag | Source |
|---|---|---|---|
| params_total | 8.19 B | verified | https://huggingface.co/api/models/Qwen/Qwen3-8B |
| params_active | 8.19 B | verified | https://huggingface.co/api/models/Qwen/Qwen3-8B |
| layers | 36 | verified | https://huggingface.co/Qwen/Qwen3-8B/blob/main/config.json |
| kv_heads | 8 | verified | https://huggingface.co/Qwen/Qwen3-8B/blob/main/config.json |
| q_heads | 32 | verified | https://huggingface.co/Qwen/Qwen3-8B/blob/main/config.json |
| head_dim | 128 | verified | https://huggingface.co/Qwen/Qwen3-8B/blob/main/config.json |
| hidden_size | 4096 | verified | https://huggingface.co/Qwen/Qwen3-8B/blob/main/config.json |
| context | 32768 tokens | sourced | https://huggingface.co/Qwen/Qwen3-8B |
| license | Apache-2.0 | verified | https://www.apache.org/licenses/LICENSE-2.0 |

## Memory and quants (total = weights + F16 KV + overhead)

| Quant | bpw | File GB | At 8k | At 32k | At 128k | KLD | 24 GB card, 8k |
|---|---|---|---|---|---|---|---|
| Q3_K_M | 3.91 | 4 [estimate] | 5.8 | 9.6 | — | 0.071 [sample] | fit |
| Q4_K_M | 4.85 | 5.03 [verified] | 6.8 | 10.6 | — | 0.017 [sample] | fit |
| Q5_K_M | 5.69 | 5.85 [verified] | 7.6 | 11.4 | — | 0.008 [sample] | fit |
| Q6_K | 6.56 | 6.73 [verified] | 8.5 | 12.3 | — | 0.004 [sample] | fit |
| Q8_0 | 8.5 | 8.71 [verified] | 10.5 | 14.3 | — | 0.001 [sample] | fit |
| F16 | 16 | 16.38 [estimate] | 18.2 | 22.0 | — | — | fit |

## Speed (default quant, 8k context, single stream)

| Device | Verdict | Decode tok/s | Flag |
|---|---|---|---|
| [GeForce RTX 5060 Ti 16 GB](/hardware/rtx-5060-ti-16g.md) | fit | 43.1 | estimate |
| [GeForce RTX 4060 Ti 16 GB](/hardware/rtx-4060-ti-16g.md) | fit | 27.7 | estimate |
| [Radeon RX 7900 XTX 24 GB](/hardware/rx-7900-xtx-24g.md) | fit | 92.3 | estimate |
| [GeForce RTX 3090 24 GB (used)](/hardware/rtx-3090-24g-used.md) | fit | 90.0 | estimate |
| [CPU only · Ryzen 9 9950X + 128 GB DDR5](/hardware/cpu-ryzen-9-9950x-128g.md) | fit | 7.2 | estimate |
| [2× RTX 3090 (used) 48 GB](/hardware/2x-rtx-3090-48g-used.md) | fit | 90.0 | estimate |
| [Mac Studio M5 Max (base)](/hardware/mac-studio-m5-max.md) | fit | 33.2 | estimate |
| [Ryzen AI Max+ 395 mini PC 128 GB](/hardware/ryzen-ai-max-395-128g.md) | fit | 18.5 | estimate |
| [GeForce RTX 4090 24 GB](/hardware/rtx-4090-24g.md) | fit | 97.0 | estimate |
| [NVIDIA DGX Spark FE 128 GB](/hardware/dgx-spark-128g.md) | fit | 19.7 | estimate |
| [MacBook Pro M4 Max 40c · 128 GB](/hardware/macbook-pro-m4-max-128g.md) | fit | 39.4 | estimate |
| [Mac Studio M5 Ultra (base)](/hardware/mac-studio-m5-ultra.md) | fit | 86.6 | estimate |
| [GeForce RTX 5090 32 GB](/hardware/rtx-5090-32g.md) | fit | 172.4 | estimate |
| [Mac Studio M5 Ultra · 512 GB](/hardware/mac-studio-m5-ultra-512g.md) | fit | 86.6 | estimate |

## Quality

LMArena text score: 1270 [sample] (Placeholder: this model has no entry in the LMArena text leaderboard dataset (checked 2026-10-12). Not a real score.)

JSON: https://thelocalbench.com/api/v1/models/qwen3-8b.json · Configurator: https://thelocalbench.com/configurator?model=qwen3-8b