Agnes 3.0 Flash Preview — Multimodal LLM AI Model
Agnes 3.0 Flash Preview — 33B hybrid-attention vision-language model, 262K context, bf16. Open-weights preview checkpoint, distinct from the production Agnes 3.0 Flash.
Details
- CategoryMultimodal LLM
- Year0
- LicenseApache-2.0
Inputs & Outputs
- TextInputstring
Text prompt or question
- Image 1Inputimageoptional
Optional reference image for visual understanding. When provided, the model analyzes the image content together with the text prompt. Supports up to 4096x4096 resolution.
- Image 2Inputimageoptional
Optional second reference image. When provided alongside Image 1, the model can compare, contrast, or jointly reason over both images with the text prompt. Supports up to 4096x4096 resolution.
- TextOutputstring
Generated text response
Parameters
- max_new_tokensintegerdefault 2048range 1–8192advanced
Max new tokens (1-8192)
Maximum number of tokens the model may generate in the response. Default 2048. Lower for terse answers / cost saving; raise (up to 8192) for long-form generation. Generation also stops at natural end-of-sequence regardless of this cap.
- temperaturefloatdefault 0.7range 0–2advanced
Temperature (0-2)
Sampling temperature. 0.0 = greedy / deterministic (best for factual Q&A, OCR, code). 0.7 (default) = balanced creativity. >1.0 = more divergent / creative. Raise for brainstorming, lower for precision tasks.
- top_pfloatdefault 0.8range 0–1advanced
Top-p (nucleus)
Nucleus sampling threshold — keeps the smallest set of tokens whose cumulative probability ≥ top_p, samples from that set. 1.0 = unrestricted, 0.8-0.95 (typical defaults) = balanced. Lower values produce more focused output. Combined with top_k.
- top_kintegerdefault 20range 1–100advanced
Top-k
Top-k sampling — restricts the candidate set to the k highest-probability tokens. 50 is permissive, 20 (Qwen default) is tighter. Lower for more deterministic output, higher for variety.
- reasoning_effortstringdefault Offoptions: Off, Low, Medium, XHighadvanced
Reasoning effort (Off/Low/Medium/XHigh)
How long Agnes thinks before answering. 'Off' replies directly and is the fastest by a wide margin — measured at 0.7s against 6.9s for 'Low' on the same prompt. The three levels run an internal chain-of-thought first, which costs latency and tokens and needs a larger max_new_tokens to reach an answer at all: at 64 tokens the model is still thinking when generation stops. Raise both together, or leave this Off for ordinary replies.
Alternatives in Multimodal Language Models
- Gemma 4 31B
Gemma 4 31B Dense flagship VLM, 256K context, thinking mode.
- Gemma4-E2B
Gemma 4 E2B compact VLM, 128K context, bf16.
- Mage-VL
Mage-VL — 4B codec-native streaming multimodal model from Microsoft. A from-scratch Mage-ViT visual encoder on a Qwen3-4B decoder, trained to read images and video through codec structure rather than uniform frame sampling. This entry serves its image-and-text mode; its video and proactive-streaming modes are not wired up here.
- Qwen3.6-35B-MoE
Qwen 3.6 35B/A3B MoE VLM, 256K context. fast inference.
- Qwen3.8-27B
Qwen 3.8 27B Dense VLM, 256K context, bf16.