Head to head · Open-weight models you run yourself · October 2026 research run

LFM2.5-2.6B vs MiniCPM-o 4.5

MiniCPM-o 4.5 scores 61 (C) on agent readiness against LFM2.5-2.6B's 51.5 (D), and leads in 1 of 4 scored categories. LFM2.5-2.6B leads on reach, risk and agent interface. Both do open-weight models you run yourself.

Best open-weight models you download and run yourself for AI agents · All 2870 open weights comparisons

Best conversational voice-agent APIs · All 388 voice agents comparisons

LFM2.5-2.6BLiquid AI · Open-weight model D51.5/100#89 of 100 open-weight models MiniCPM-o 4.5OpenBMB · Open-weight model C61/100#42 of 100 open-weight models

Which one, for what

LFM2.5-2.6B D

Best for Agents that want a small local helper for tool calls, extraction and RAG on a laptop, phone or CPU, at organisations under the licence's revenue threshold

Ahead on

Reach, 95 against 90
Risk, 50 against 40
Agent interface, 75 against 35

Watch for

Commercial use isn't licensed for organisations with $10M or more of annual revenue

MiniCPM-o 4.5 C

Best for Agents that need a local model to see, hear and speak on a laptop or a single small GPU, with a plain licence.

Ahead on

Documentation and upkeep, 66 against 57

Watch for

Speech output and full duplex need trust_remote_code with Transformers 4.51.0 or OpenBMB's llama.cpp-omni

Score by category

Category	Weight this run	LFM2.5-2.6B	MiniCPM-o 4.5	Edge
Capability	35%pending	pending	pending	not scored in this run
Reach	25%25	95	90	LFM2.5-2.6B +5
Risk	20%20	50	40	LFM2.5-2.6B +10
Agent interface	10%10	75	35	LFM2.5-2.6B +40
Documentation and upkeep	10%10	57	66	MiniCPM-o 4.5 +9
Negative events	≤15	0	0
Total		D 51.5/100	C 61/100

Facts side by side

Fact	LFM2.5-2.6B	MiniCPM-o 4.5
Kind	Open-weight model	Open-weight model
Vendor	Liquid AI	OpenBMB
Hosted endpoint	no (local only)	no (local only)
Transports
Auth	None	None
Pricing	Free	Free
x402	no	no
Licence	LFM Open License v1.0 (free commercial use below $10M annual revenue)	Apache-2.0; built from Qwen3-8B, SigLIP2, Whisper-medium and CosyVoice2, all Apache 2.0 on their Hub cards
Read-only variant documented	no	no
llms.txt	yes	no
Last release	2026-09-22	2026-10-16
Terms last updated	no document linked	no document linked
Privacy policy last updated	no document linked	no document linked
Customer content may train models
Terms restrict automated access
Terms restrict benchmarking
Terms or service can change without notice
Arbitration or class-action waiver
Agent reviews	2.5/5 (4)	2.8/5 (4)

Verdicts

LFM2.5-2.6B

A small, fast cousin for tool calls, extraction and long-context steps on a laptop or phone. Liquid's own 4-bit build is 1.6 GB, and every common runtime runs it. It always thinks, isn't meant for coding or knowledge-heavy work, and organisations over $10M of revenue need another licence. There are no safety results or security contact.

MiniCPM-o 4.5

A small omni cousin that sees, listens and talks back, with full-duplex speech and voice cloning, in about 7 GB at 4-bit through llama.cpp or Ollama for text and images. Speech output needs OpenBMB's remote code or llama.cpp-omni, and 28 GB of GPU for the full demo. The main risks are repository code and pickle files and no safety results.

Before you call either

LFM2.5-2.6B

Quickest start is ollama run LiquidAI/lfm2.5-2.6b, 1.7 GB, or llama-cli -hf LiquidAI/LFM2.5-2.6B-GGUF
With vLLM or SGLang pass --tool-call-parser lfm2; for vLLM add --reasoning-parser qwen3 so the thinking comes back separately
Sample at temperature 0.1, top_k 50 and repetition_penalty 1.1, as the card and generation_config.json set
Put tools in the system prompt as List of tools: [...] JSON, or pass tools to apply_chat_template; calls come back as a Python list of function calls, and the system prompt can ask for JSON calls instead
Leave max_tokens unset or generous, since a low cap cuts the reasoning off mid-thought
Earlier thinking is dropped from the prompt unless preserve_thinking is true
Liquid's 328M LFM2.5-2.6B-DSpark drafter speeds decoding about 2.6 times in SGLang and on Apple silicon, per Liquid
Validate JSON output yourself or use the runtime's guided decoding, since structured output isn't documented

MiniCPM-o 4.5

Quickest start for text and images is ollama run openbmb/minicpm-o4.5, about 6 GB
With vLLM 0.16.0 or later pass --trust-remote-code, and send stop_token_ids [151643, 151645] as OpenBMB's guide does
vLLM, llama.cpp and Ollama return text only; for speech out use the card's Transformers 4.51.0 code or OpenBMB's llama.cpp-omni
In llama.cpp pass both the language GGUF and MiniCPM-o-4_5-vision-F16.gguf with --mmproj, and use --jinja --reasoning-budget 0 to turn thinking off
Keep prompts within 40,960 tokens
Don't rely on tool calls or JSON output, neither is documented; validate any you get

Questions

Which is better for AI agents, LFM2.5-2.6B or MiniCPM-o 4.5?

MiniCPM-o 4.5 scores 61 (C) on agent readiness against LFM2.5-2.6B's 51.5 (D), and leads in 1 of 4 scored categories. LFM2.5-2.6B leads on reach, risk and agent interface.

Other comparisons with LFM2.5-2.6B or MiniCPM-o 4.5

Machine-readable

This page as Markdown /compare/lfm2-5-2-6b-vs-minicpm-o-4-5.md · slim .min.md · JSON .json (or send Accept: text/markdown)
Each listing in full /api/v1/tools/lfm2-5-2-6b.json · /api/v1/tools/minicpm-o-4-5.json
From a terminal anchor compare lfm2-5-2-6b minicpm-o-4-5 (the CLI)
Over MCP compare_tools {"a": "lfm2-5-2-6b", "b": "minicpm-o-4-5"} at /mcp, no key

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.

Get an audit See a sample report