Step3 VL 10B
Files and memory at 8K context
| Quant | File | Files | + KV + runtime | Smallest fit | Source |
|---|---|---|---|---|---|
| Q4_K_M JamePeng2023/Step3-VL-10B-GGUF | 4.68 GiB | 1 | 6.8 GiB | 8 GB GPU | HF files |
| Q5_K_M Vastined/Step3-VL-10B-GGUF | 5.45 GiB | 1 | 7.6 GiB | 12 GB GPU | HF files |
| Q8_0 JamePeng2023/Step3-VL-10B-GGUF | 8.11 GiB | 1 | 10.2 GiB | 12 GB GPU | HF files |
| FP16 seanbailey518/Step3-VL-10B-GGUF | 15.26 GiB | 1 | 17.4 GiB | 20 GB GPU | HF files |
No verified file: MLX 4-bit, MLX 8-bit. Sizes are exact file bytes (text weights; projectors listed separately). “+” = recurrent state or other terms unknown. Smallest fit adds 0.5 GiB spare; Macs use 75% of memory.
Quality
Artificial Analysis Intelligence Index
7.7 AA estimate
Variant: Step3 VL 10B · observed 2026-10-17 · source
Rank #73 of 75 researched 2026 models.
Arena text rating
Unknown
Scores describe the evaluated hosted variant, not each quantized file. AA and Arena are never averaged.
Reported speeds 0
No reported measurement yet. Submit one.
Community threads
- tool calls sometimes fail on vllm Hugging Face · 2026-01-24
- Not support flash_attn_2？ Hugging Face · 2026-01-22
Qualitative signals only; titles are not verified claims.