Agnes 3.0 Flash — Hardware Requirements & GPU Compatibility
VisionReasoning
Agnes-3.0-Flash Preview is Agnes AI's 33-billion-parameter open-weight multimodal model offering competitive reasoning, coding, and instruction-following at lower hardware cost than flagship-class models; it understands text, images, and video, supports tool calling, and lets developers dial reasoning effort up or down. This open-weight Preview checkpoint is distinct from the larger production/API "Agnes 3.0 Flash" model on third-party leaderboards, which uses a different checkpoint with a 1-million-token context window, so its benchmark results don't transfer to these weights. Architecturally it is a hybrid-attention decoder where three of every four layers use a recurrent gated-delta-rule mechanism instead of standard attention, so only 18 of its 72 layers hold a KV cache that grows with context. At 33 billion dense parameters, it needs a high-end consumer GPU or multi-GPU setup once quantized. Context length is 262,144 tokens. It is released under the Apache 2.0 license, permitting unrestricted commercial and research use, and was published in September 2026.
1.9K downloads
244 likesSpecifications
- Publisher
- Agnes-AI
- Parameters
- 33.1B
- Architecture
- AgnesForConditionalGeneration
- Context Length
- 262,144 tokens
- Vocabulary Size
- 248,320
- Release Date
- 2026-09-11
- License
- Apache 2.0
Get Started
HuggingFace
How Much VRAM Does Agnes 3.0 Flash Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| Q2_Kest. | 3.40 | 14.9 GB | 78.8 GB | 14.06 GB | 2-bit quantization with K-quant improvements |
| Q3_K_Mest. | 3.90 | 16.9 GB | 80.9 GB | 16.13 GB | 3-bit medium quantization |
| IQ4_XS | 4.30 | 18.6 GB | 82.5 GB | 17.79 GB | Importance-weighted 4-bit, compact |
| Q4_K_M | 4.80 | 20.7 GB | 84.6 GB | 19.85 GB | 4-bit medium quantization — most popular sweet spot |
| Q5_K_M | 5.70 | 24.4 GB | 88.3 GB | 23.58 GB | 5-bit medium quantization — good quality/size tradeoff |
| Q6_K | 6.60 | 28.1 GB | 92.0 GB | 27.30 GB | 6-bit quantization, very good quality |
| Q8_0 | 8.00 | 33.9 GB | 97.8 GB | 33.09 GB | 8-bit quantization, near-lossless |
| BF16est. | 16.00 | 67.0 GB | 130.9 GB | 66.18 GB | Brain floating point 16 — preferred for training |
est.= calculated VRAM estimate; no published GGUF file found for that quantization yet. Other rows are verified against real community uploads.
Which GPUs Can Run Agnes 3.0 Flash?
Q4_K_M · 20.7 GBAgnes 3.0 Flash (Q4_K_M) requires 20.7 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 27+ GB is recommended. Using the full 262K context window can add up to 63.9 GB, bringing total usage to 84.6 GB. 7 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti.
Runs great
— Plenty of headroom
Which Devices Can Run Agnes 3.0 Flash?
Q4_K_M · 20.7 GB41 devices with unified memory can run Agnes 3.0 Flash, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, Mac Mini M4 Pro (24 GB).
Runs great
— Plenty of headroomDecent
— Enough memory, may be tightWhere to Download Agnes 3.0 Flash
Community quantizations of this model — GGUF for llama.cpp, Ollama, and LM Studio, plus AWQ/MLX variants where available.
Frequently Asked Questions
- How much VRAM does Agnes 3.0 Flash need?
Agnes 3.0 Flash requires 20.7 GB of VRAM at Q4_K_M, or 67.0 GB at BF16. Full 262K context adds up to 63.9 GB (84.6 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 33.1B × 4.8 bits ÷ 8 = 19.9 GB
KV Cache + Overhead ≈ 0.8 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 64.7 GB (at full 262K context)
VRAM usage by quantization
Q4_K_M20.7 GBQ4_K_M + full context84.6 GB- Can NVIDIA GeForce RTX 4090 run Agnes 3.0 Flash?
Yes, at Q4_K_M (20.7 GB) or lower. Higher quantizations like Q5_K_M (24.4 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.
- What's the best quantization for Agnes 3.0 Flash?
For Agnes 3.0 Flash, Q4_K_M (20.7 GB) offers the best balance of quality and VRAM usage. Q5_K_M (24.4 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 14.9 GB.
VRAM requirement by quantization
Q2_K14.9 GBIQ4_XS18.6 GBQ4_K_M ★20.7 GBQ5_K_M24.4 GBQ6_K28.1 GBBF1667.0 GB★ Recommended — best balance of quality and VRAM usage.
- Can I run Agnes 3.0 Flash on a Mac?
Agnes 3.0 Flash requires at least 14.9 GB at Q2_K, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Agnes 3.0 Flash locally?
Yes — Agnes 3.0 Flash can run locally on consumer hardware. At Q4_K_M quantization it needs 20.7 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Agnes 3.0 Flash?
At Q4_K_M, Agnes 3.0 Flash can reach ~232 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~32 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B200 → 8000 ÷ 20.7 × 0.65 = ~252 tok/s
Estimated speed at Q4_K_M (20.7 GB)
~252 tok/s~32 tok/s~252 tok/s~232 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Agnes 3.0 Flash?
At Q4_K_M, the download is about 19.85 GB. The full-precision BF16 version is 66.18 GB. The smallest option (Q2_K) is 14.06 GB.
- Which GPUs can run Agnes 3.0 Flash?
7 consumer GPUs can run Agnes 3.0 Flash at Q4_K_M (20.7 GB). Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090. 1 GPU have plenty of headroom for comfortable inference.
- Which devices can run Agnes 3.0 Flash?
41 devices with unified memory can run Agnes 3.0 Flash at Q4_K_M (20.7 GB), including AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10, Asus ROG Flow Z13 (2025, Ryzen AI Max+ 395, 128 GB), Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.