Best AI Models for Intel Arc Pro B65 (32.0GB)
B60-class compute with B70 memory — 32 GB at 608 GB/s for about $910, so capacity per dollar is its whole argument. Add-in-board partners only; Intel ships no reference card and set no MSRP.
32 GB positions this hardware in the professional tier for local AI. Most popular open-source models run comfortably, and even large 70B parameter models are accessible at lower quantization levels.
This memory amount is a sweet spot for enthusiasts and professionals. You can run 13B–30B models like DeepSeek R1 Distill at Q5 or Q6 quality with smooth token generation, and 7B models at near-lossless precision. The 70B class of models (Llama 3 70B, Qwen 72B) becomes possible at Q2–Q3 quantization, though with some quality trade-off. For day-to-day use with coding assistants, chat models, and reasoning tasks, this tier delivers an excellent experience.
Runs Well
- 7B–13B models at Q6–Q8 quality
- 14B–30B models at Q4–Q5 quality
- Small models (3B–7B) at FP16 precision
- Vision-language models at good quality
Challenging
- 70B models only at Q2–Q3 (noticeable quality loss)
- Large context windows with 30B+ models
What LLMs Can Intel Arc Pro B65 Run?
254 models · 91 excellent · 93 good
Showing compatibility for Intel Arc Pro B65
| Model | Quant | VRAM | Speed | Context | Status | Grade |
|---|---|---|---|---|---|---|
Q4_K_M·70.3 t/s tok/s·262K ctx·RUNS GREAT | Q4_K_M | 16.1 GB | 70.3 t/s | 262K | RUNS GREAT | S86 |
Q4_K_M·17.5 t/s tok/s·262K ctx·DECENT | Q4_K_M | 17.4 GB | 17.5 t/s | 262K | DECENT | B57 |
Q4_K_M·16.6 t/s tok/s·131K ctx·DECENT | Q4_K_M | 18.3 GB | 16.6 t/s | 131K | DECENT | B56 |
Q4_K_M·63.1 t/s tok/s·262K ctx·RUNS WELL | Q4_K_M | 21.4 GB | 63.1 t/s | 262K | RUNS WELL | A84 |
Q4_K_M·60.8 t/s tok/s·262K ctx·RUNS WELL | Q4_K_M | 19.3 GB | 60.8 t/s | 262K | RUNS WELL | A84 |
Q4_K_M·66.9 t/s tok/s·262K ctx·RUNS GREAT | Q4_K_M | 21.2 GB | 66.9 t/s | 262K | RUNS GREAT | S85 |
Q4_K_M·70.3 t/s tok/s·262K ctx·RUNS GREAT | Q4_K_M | 16.1 GB | 70.3 t/s | 262K | RUNS GREAT | S86 |
Q4_K_M·57.0 t/s tok/s·262K ctx·RUNS WELL | Q4_K_M | 21.9 GB | 57.0 t/s | 262K | RUNS WELL | A82 |
Q4_K_M·52.6 t/s tok/s·203K ctx·RUNS WELL | Q4_K_M | 19.8 GB | 52.6 t/s | 203K | RUNS WELL | A81 |
Q4_K_M·60.7 t/s tok/s·500K ctx·RUNS WELL | Q4_K_M | 18.7 GB | 60.7 t/s | 500K | RUNS WELL | A84 |
Q4_K_M·63.1 t/s tok/s·262K ctx·RUNS WELL | Q4_K_M | 21.4 GB | 63.1 t/s | 262K | RUNS WELL | A84 |
Q4_K_M·14.9 t/s tok/s·262K ctx·DECENT | Q4_K_M | 20.4 GB | 14.9 t/s | 262K | DECENT | B55 |
Q4_K_M·141.3 t/s tok/s·128K ctx·RUNS GREAT | Q4_K_M | 5.5 GB | 141.3 t/s | 128K | RUNS GREAT | S96 |
Q4_K_M·47.8 t/s tok/s·262K ctx·RUNS WELL | Q4_K_M | 6.4 GB | 47.8 t/s | 262K | RUNS WELL | A79 |
Q4_K_M·61.1 t/s tok/s·262K ctx·RUNS WELL | Q4_K_M | 18.7 GB | 61.1 t/s | 262K | RUNS WELL | A84 |
Q4_K_M·93.8 t/s tok/s·131K ctx·RUNS GREAT | Q4_K_M | 7.7 GB | 93.8 t/s | 131K | RUNS GREAT | S90 |
Intel Arc Pro B65 Specifications
- Brand
- Intel
- Architecture
- Xe2 (Battlemage), BMG-G31
- VRAM
- 32.0 GB GDDR6
- Memory Bandwidth
- 608.0 GB/s
- Tensor Cores
- 160
- FP16 Performance
- 98.50 TFLOPS
- TDP
- 200W
- Release Date
- 2026-03-25
- MSRP
- $910
Get Started
Prompt Processing
Estimated for Mellum2 12B A2.5B Base, the compute-bound phase that reads your prompt before the first reply token appears.
Short chat
141 ms
512 tok prompt
Long chat
1.1 s
4,096 tok prompt
Document / codebase
9.0 s
32,768 tok prompt
Prefill is compute-bound and a different number from the decode tok/s shown elsewhere on this page — how prompt processing works →
Efficiency & Value
Based on Mellum2 12B A2.5B Base at ~45.7 tok/s decode.
Tokens per watt
Higher is better.
$ per tok/s (MSRP)
MSRP-based, not street price. Lower is better.
Performance figures are estimates calibrated as of 2026-09-21 — see calibration basis →
GPUs to Consider Over Intel Arc Pro B65
Similar GPUs and upgrades with more VRAM or higher bandwidth for AI
NVIDIA GeForce RTX 5090
NVIDIA · Blackwell
AMD Instinct MI210
AMD · CDNA 2
NVIDIA A100 40GB PCIe
NVIDIA · Ampere
NVIDIA RTX 6000 Ada Generation
NVIDIA · Ada Lovelace
NVIDIA RTX PRO 4500 Blackwell
NVIDIA · Blackwell
AMD Radeon PRO W7900
AMD · RDNA 3
Frequently Asked Questions
- Can Intel Arc Pro B65 run Gemma 4 26B A4B IT?
Yes, the Intel Arc Pro B65 with 32 GB can run Gemma 4 26B A4B IT, Qwen3.8 27B, Muse Glimmer 30B, and 2382 other models. 1087 models run at excellent quality, and 849 at good quality. Check the compatibility table above for the full list with VRAM usage and estimated speed.
- Is Intel Arc Pro B65 good for AI?
The Intel Arc Pro B65 has 32 GB of GDDR6, making it excellent for running local AI models. It supports 1936 models at good quality or better. With 608.0 GB/s memory bandwidth, it delivers solid token generation speeds. This is an enthusiast-grade GPU that handles most popular open-source LLMs.
- How many parameters can Intel Arc Pro B65 handle?
With 32 GB, the Intel Arc Pro B65 supports models from 3B to 30B parameters depending on quantization level. At Q4_K_M (the recommended sweet spot), you can fit roughly 53B parameters. This means 7B models at high quality (Q6/Q8) or 30B+ models at Q4.
- What quantization should I use on Intel Arc Pro B65?
For the best balance of quality and speed on the Intel Arc Pro B65, start with Q4_K_M — it preserves ~85% of the original model quality while keeping VRAM usage reasonable. With 24+ GB, you have the headroom to run 7B models at Q5_K_M or even Q6_K for noticeably better output quality. For larger 30B models, Q4_K_M remains the sweet spot.
- How fast is Intel Arc Pro B65 for AI inference?
With 608.0 GB/s memory bandwidth, the Intel Arc Pro B65 achieves approximately 68 tokens/sec on a 7B model at Q4_K_M — that's very fast, well above conversational speed. A 14B model runs at ~34 tok/s. Token generation speed scales inversely with model size — smaller models are significantly faster.
tok/s = (608 GB/s ÷ model GB) × efficiency
Smaller models = faster inference. Memory bandwidth is the main bottleneck for token generation speed.
Estimated speed on Intel Arc Pro B65
~70 tok/s~18 tok/s~17 tok/s~63 tok/sReal-world results typically within ±20%. Speed depends on quantization kernel, batch size, and software stack.
- What's the best model for Intel Arc Pro B65?
The top-rated models for the Intel Arc Pro B65 are Gemma 4 26B A4B IT, Qwen3.8 27B, Muse Glimmer 30B. The best choice depends on your use case: coding assistants benefit from code-tuned models, while general chat works well with instruction-tuned models like Llama or Qwen.
- What power supply and cooling does Intel Arc Pro B65 need?
The Intel Arc Pro B65 has a TDP of 200 W. A good rule of thumb is to provide at least double the GPU's TDP to cover the rest of the system — that means a 550 W PSU or larger. A mid-tower case with one intake and one rear exhaust is usually sufficient. Keep dust filters clean, as sustained inference generates continuous heat rather than the brief spikes typical of gaming.
- Anything to watch out for with Intel Arc Pro B65?
B60-class compute with B70 memory — 32 GB at 608 GB/s for about $910, so capacity per dollar is its whole argument. Add-in-board partners only; Intel ships no reference card and set no MSRP.