whatmodel.sh / MiMo-V2.5-Pro (1T-A42B)

MiMo-V2.5-Pro (1T-A42B)

+ contribute

MiMo-V2.5-Pro (1T-A42B) is an open-weights language model from Xiaomi, released April 2026. Its published size is 1023B parameters (42B active per token) and a 1M-token context window. Licence: mit.

69.7±1.0#87 of 719 LLMsfrom 21 benchmark results across 6 sources

$ how it compares

The LLMs ranked either side of MiMo-V2.5-Pro (1T-A42B), out of 828 with a score, then more from Xiaomi.

#	model	score	where	RAM	released
93	GLM-5 Z.ai	70±1	api $1.55/M +local	~119GB	2026-02
94	Qwen3-Max (2026-01-23) Alibaba	70±1	comparison only		2026-01
95	MiMo-V2.5-Pro (1T-A42B) Xiaomi	70±1	api $0.65/M +local	~163GB	2026-04
96	Inkling (975B-A41B) Thinking Machines	70±1	api $1.73/M +local	~156GB	2026-07
97	Muse Glimmer 30B Meta	70±1	api $0.64/M +local	~19GB	2026-08
	more from Xiaomi
15	MiMo-V2.6-Pro Xiaomi	79±1	api $0.54/M +local	~443GB	2026-09
27	MiMo-V2.6-Flash (309B-A15B) Xiaomi	77±1	api $0.17/M +local	~136GB	2026-09
91	MiMo-V2-Pro Xiaomi	70±1	comparison only		2026-03

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #75 of 300 · 59% of its score · 13 tests

Beats GPT-5, Gemini 3 Flash (Preview), Grok 4.20 (0309 reasoning) here, though they rank higher overall. Trails MiMo-V2.5 (310B-A15B), Qwen3.6 27B, Muse Glimmer 30B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Arena Agent (October 2026 aggregate IPS)	-0.0752 IPS ✎	▁▁▁▄▂▃▇█▂▂▁▃ beats 6/49	avg -0.312 IPS	▇▇▇▇▇▇▇ 17%	sharp
Terminal-Bench 2.1 lab	65.2% ✎	▁·▁·▁▁▄▃▄█▃▇ beats 30/56	hard · avg 21%	▇▇▇ 6.3%	at its level · small run
SWE-bench Pro (split unspecified) lab	57.2% ✎	▃·▁▁▁▂▂▃█▇▃▁ beats 35/55	very hard · avg 9.8%	▇▇ 5.5%	sharp · at its level · small run
GameLogicBench solve rate (72 Godot game-logic tasks, Claude Code scaffold)	16.7% ✎	█▄·█·▄·██▄·▄ beats 1/11	very hard · avg 4.7%	▇▇ 5.4%	typical
SWE-bench Verified lab	78.9% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 122/131	hard · avg 26%	▇▇ 5.2%	sharp · small run
Terminal-Bench 4.0 lab	1.5% ✎	▁·▃··▁▇▇▅██▃ beats 1/30	very hard · avg 0.4%	▇▇ 5.0%	sharp · small run
Terminal-Bench 2.0 lab	68.4% ✎	▂▂▅▅████▆▃▃▃ beats 32/36	very hard · avg 15%	▇▇ 4.0%	at its level · small run
EEGAgentBench overall (tool-based EEG analysis agents, 6 tasks)	49.1% ✎	▂·▁▁··▇▇█▂·▁ beats 16/25	hard · avg 29%	▇ 2.5%	blunt · at its level

5 more tests

test	result	vs the others in its run	difficulty	counts	why
Humanity's Last Exam with tools lab	48.0% ✎	▂▂▂▁▄▂▃▆█▇▂▆ beats 26/54	hard · avg 15%	▇ 2.2%	at its level · small run
Agents' Last Exam (metric/harness unspecified) lab	13.2% ✎	▃·▂·█▂···▂▄▄ beats 0/11	very hard · avg 6.4%	▇ 2.0%	small run
Toolathlon-Verified lab	49.1% ✎	▂·▂▂▅▃···▂██ beats 5/18	hard · avg 31%	▇ 1.6%	blunt · at its level · small run
τ³-bench lab	72.9% ✎	▁▁▂▁·▁▁·▁▁▂█ beats 23/25	medium · avg 36%	▇ 1.5%	blunt · at its level · small run
DeepSWE v1.1 (resolved) lab	19.0% ✎	▃····▁▁▁▆██▆ beats 1/28	hard · avg 18%	▇ 1.1%	at its level · noisy · small run

coding▁▁▁▃▆██▇▆▄▃▁ #58 of 458 · 32% of its score · 5 tests

Beats GPT-5.4, DeepSeek V4 Flash Vision (Exp), Gemini 3.1 Pro (Preview) here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
Arena WebDev (Code Arena)	1477.6 ✎	▁▂▄▃▅█▄▇▆▄▁▁ beats 65/119	avg 981	▇▇▇▇▇▇ 15%	sharp
SWE-bench Pro (split unspecified) lab	57.2% ✎	▃·▁▁▁▂▂▃█▇▃▁ beats 35/55	very hard · avg 9.8%	▇▇ 5.5%	sharp · at its level · small run
GameLogicBench solve rate (72 Godot game-logic tasks, Claude Code scaffold)	16.7% ✎	█▄·█·▄·██▄·▄ beats 1/11	very hard · avg 4.7%	▇▇ 5.4%	typical
SWE-bench Verified lab	78.9% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 122/131	hard · avg 26%	▇▇ 5.2%	sharp · small run
DeepSWE v1.1 (resolved) lab	19.0% ✎	▃····▁▁▁▆██▆ beats 1/28	hard · avg 18%	▇ 1.1%	at its level · noisy · small run

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #35 of 393 * · 8.4% of its score · 1 test

Beats GPT-6 Sol, GPT-5.6 Terra, MiMo-V2.6-Flash (309B-A15B) here, though they rank higher overall.

test	result	vs the others in its run	difficulty	counts	why
LMArena (text)	1467.5 ✎	▁▁▁▂▃▃▃▆▆▇█▃ beats 339/379	avg 1268	▇▇▇ 8.4%	blunt

web search······█····· #1 of 1 * · 2.2% of its score · 1 test

test	result	vs the others in its run	difficulty	counts	why
Humanity's Last Exam with tools lab	48.0% ✎	▂▂▂▁▄▂▃▆█▇▂▆ beats 26/54	hard · avg 15%	▇ 2.2%	at its level · small run

other tests6 not in a task · 17% of its score

test	result	vs the others in its run	difficulty	counts	why
WildClawBench	43.0% ✎	▃·▅▅█·▅█▅▃▃▃ beats 11/17	very hard · avg 11%	▇▇ 4.6%	at its level
Humanity's Last Exam lab	34.0% ✎	▁▁▁▃▆▆▇██▆▁▃ beats 79/104	very hard · avg 5.8%	▇ 3.5%	at its level · small run
AutomationBench v1.0.6 lab	16.0% ✎	▃····▃·▃▃··█ beats 1/6	very hard · avg 5.6%	▇ 2.9%	at its level · small run
SEC-Bench Pro (pass@1) lab	17.7% ✎	▄··▄···▄·▄·█ beats 0/5	very hard · avg 6.4%	▇ 2.6%	at its level · small run
Claw-Eval Pass^3 lab	63.8% ✎	█▄▄·▄·▄▄·▄·▄ beats 7/8	hard · avg 27%	▇ 1.9%	at its level · small run
ExploitGym (pass@1) lab	0.2% ✎	▄···▄··▄▄··█ beats 0/5	very hard · avg 0.6%	▇ 1.4%	too hard for it · noisy · small run

$ where to run it

Through an API

api · Novita AI · $0.65/M · $0.522 in / $1.044 out per 1M tokens · 1M context · novita.ai ↗ · price sourceUSD per 1M tokens, Novita AI serverless chat completions, model xiaomimimo/mimo-v2.5-pro (status active; 1M context): input $0.522, output $1.044; cache read excluded. Xiaomi's own API lists mimo-v2.5-pro at $0.435/$0.87 (overseas, cache miss) but marks it 'to be deprecated' on 2026-10-21. Currency: USD. Novita AI official pricing page, observed 2026-09-25.

llama.cpp · recommended

version

Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
Windows
irm https://llama.app/install.ps1 | iex
other ways to install ↗
Run MiMo-V2.5-Pro (1T-A42B) · needs ~614.3 GB · from huggingface.co/unsloth
chat in the terminal
llama cli -hf unsloth/MiMo-V2.5-Pro-GGUF:UD-Q4_K_XL
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf unsloth/MiMo-V2.5-Pro-GGUF:UD-Q4_K_XL

LM Studio

version

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run MiMo-V2.5-Pro (1T-A42B) · needs ~614.3 GB · from huggingface.co/unsloth
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/unsloth/MiMo-V2.5-Pro-GGUF@UD-Q4_K_XL -y
Then run lms ls to see its name, and lms chat followed by that name

Weights and documentation: huggingface.co/XiaomiMiMo ↗

$ how the score was worked out · LMArena leaderboard dataset 40%, XiaomiMiMo/MiMo-V2.5-Pro mode… 24%, Xiaomi MiMo official model ca… 23% and 3 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
LMArena leaderboard dataset	3	▇▇▇▇▇▇▇▇▇▇▇▇ 40%	CC BY 4.0
XiaomiMiMo/MiMo-V2.5-Pro model card · lab-reported	7	▇▇▇▇▇▇▇ 24%	MIT
Xiaomi MiMo official model cards and documentation · lab-reported	8	▇▇▇▇▇▇▇ 23%	Factual first-party specification (not licensed content)
GameLogicBench: Evaluating Coding Agents on Runtime Game Logic with Tick-Level State Assertions (arXiv:2609.21562v2)	1	▇▇ 5%	CC BY 4.0
WildClawBench: A Benchmark for Real-World, Long-Horizon Agent Evaluation (arXiv:2605.10912v1)	1	▇ 5%	CC BY 4.0
EEGAgentBench: Benchmarking LLM Agents on Short- and Long-Horizon EEG Analysis (arXiv:2609.31632v1)	1	▇ 2%	CC BY 4.0
prior · pull toward the middle		▇ 0%
weighted average		= 69.7

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-10-11 02:20 UTC · sources and licences · json