As the frontier of foundation models pushes toward ever-larger parameter counts and ever-longer context windows, the question is no longer just how much we can scale, but how efficiently we can do so.
Agnes-2.5-Pro-Theta
Agnes-AI·Instruct·released 15 Nov 2026
hf download Agnes-AI/Agnes-2.5-Pro-ThetaAgnes 2.5 Pro Theta is a multimodal mixture-of-experts checkpoint distributed with its model implementation, tokenizer, chat template, and image/video processors. This release provides BF16 weights for local inference through Transformers.
How much VRAM does Agnes-2.5-Pro-Theta need?
At Q4_K_M Agnes-2.5-Pro-Theta needs about 117 GB with an 8K context (117 GB at Q4, 373 GB at full precision). Pick your GPU to see what fits and how fast it runs.
Needs: ≈117 GB · Q4_K_M · 8K context — Pick a GPU to check the fit.
Advanced: context 8K · KV cache 16-bit
Smallest GPU for each quant and context (KV cache 16-bit)
| Quantisation | 4K | 8K | 16K | 32K | 64K | 128K |
|---|---|---|---|---|---|---|
| Q3_K_M | ||||||
| Q4_K_M | ||||||
| Q5_K_M | ||||||
| Q6_K | ||||||
| Q8_0 | ||||||
| FP16 |
≤16 GB card24–32 GB48–96 GBmore
| Quantisation | Total |
|---|---|
| Q2_KHeavy lossSmallest, noticeably worse quality | ≈66 GB |
| Q3_K_MNoticeable lossSmall, some quality loss | ≈95 GB |
| Q4_K_MrecommendedSmall lossRecommended default for local use | ≈117 GB |
| Q5_K_MNear-losslessNear-lossless for most tasks | ≈136 GB |
| Q6_KNear-losslessPractically lossless | ≈157 GB |
| Q8_0LosslessLossless in practice, twice the size of Q4 | ≈201 GB |
| FP16 / BF16LosslessFull precision, as released | ≈373 GB |
Estimates for 8K tokens of context: weights + KV cache + runtime overhead. Speed assumes memory-bound decoding at your GPU’s bandwidth. Pick a GPU above and the fit column checks every row against it.
How to pick a quant
Take the biggest quant that fits with room left for context. Q4_K_M is the usual sweet spot. Below Q4 quality drops fast, above Q6 you mostly pay memory for nothing. Big models (70B+, large MoE) tolerate low quants better than small ones.
- LosslessF16, Q8_0
- Near-losslessQ6_K, Q5_K_M
- Small lossQ5_K_S, Q4_K_M, IQ4_NL, IQ4_XS, Q4_K_S
- Noticeable lossQ4_0, Q3_K_L, Q3_K_M, IQ3_M
- Heavy lossIQ3_XXS, Q3_K_S, IQ2_M, Q2_K
Typical quality loss, not measured per model. Tiers come from KL divergence against FP16 on Llama 3 8B: llama.cpp perplexity README: LLaMA 3 8B scoreboard (KLD and ΔPPL vs FP16, Wikitext-2); llama.cpp quantize tool: ΔPPL per quant type @ Llama-3-8B
How much context fits?
At 8K context, Q4_K_M needs ≈117 GB(106 GB weights + 5.0 GB KV cache + 5.8 GB overhead). Pick a GPU above to see how much context fits.
Weights + overheadKV cache
Can the Apple M2 Ultra (192 GB) run Agnes-2.5-Pro-Theta? →
Speed vs quality per quant
Smaller quants run faster and fit in less memory, but lose quality.
Fits in VRAMTight fitCPU offload✕ won't runQuality is typical KL divergence vs FP16, not measured for this model.
Real-world speed of Agnes-2.5-Pro-Theta
Tokens per second measured by people running it at home. A GPU shows a median once 3 reports agree on the same setup. Reports far from our estimate wait for a review. Measure yours with llama-bench →
No medians yet. Be one of the first 3 to report a setup.
All reports
Run Agnes-2.5-Pro-Theta locally
These commands load the original weights from Hugging Face.
from transformers import pipeline
pipe = pipeline("image-text-to-text", model="Agnes-AI/Agnes-2.5-Pro-Theta", dtype="auto", device_map="auto")
messages = [
{
"role": "user",
"content": [
{"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
{"type": "text", "text": "Describe this image."},
],
}
]
out = pipe(text=messages, max_new_tokens=256)
print(out[0]["generated_text"][-1]["content"])About Agnes-2.5-Pro-Theta
Agnes 2.5 Pro Theta is a multimodal mixture-of-experts checkpoint distributed with its model implementation, tokenizer, chat template, and image/video processors. This release provides BF16 weights for local inference through Transformers. The supplied evaluation snapshot covers eight benchmarks, with 92.9 on GPQA Diamond, 90.5 on LiveCodeBench v6, and 84.7 on AA-LCR. The package has approximately 187.8B total parameters and a configured context limit of 262,144 tokens.
Full model card on Hugging Face →Alternatives to Agnes-2.5-Pro-Theta
All alternatives →👋 Join our WeChat or Discord community. 📖 Check out the GLM-5.3-Flash blog and GLM-5 Technical report . 📍 Use GLM-5.3-Flash API services on Z.ai API Platform.
MiniMax-M3.1-Flash is a native multimodal mixture-of-experts model for coding, agentic workflows, computer use, and cybersecurity. It supports text, image, and video inputs with a context window of up to 1M tokens.
Inkling-Small is a general-purpose multimodal model that accepts text, image and audio inputs and generates text outputs. It is intended for use in English and other languages, and across multiple coding languages.
Over recent months, we have intensified our focus on developing foundation models that deliver exceptional utility and performance.
Meet Qwen3-VL — the most powerful vision-language model in the Qwen series to date.
Agnes-2.5-Pro-Theta FAQ
- How much VRAM does Agnes-2.5-Pro-Theta need?
- About 117 GB at Q4_K_M with an 8K context. Q8_0 needs about 201 GB and full precision about 373 GB. Longer contexts need more memory for the KV cache.
- Can Agnes-2.5-Pro-Theta run on the Apple M2 Ultra (192 GB)?
- Yes. It fits at Q5_K_M (about 136 GB), at an estimated 3.3 tokens/s. It fits comfortably on the Apple M3 Ultra (512 GB).
- Can I use Agnes-2.5-Pro-Theta commercially?
- Yes. It is released under Apache 2.0. Permissive: use, modify and sell freely; keep the notice and licence text.
- How do I run Agnes-2.5-Pro-Theta locally?
- It runs with Transformers. Each has copy-paste examples in its docs.
Rate this model
Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.
0 comments